JP2018174418A - Image processing apparatus and computer program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect an edge of a character that can be expressed in various forms. Luminance image data is acquired using input image data. Using the luminance image data, it is determined whether or not the target pixel is included in the first type specific area that is darker than the second type specific area. Using the luminance image data, it is determined whether or not the target pixel is included in the second type specific area that is brighter than the first type specific area. Smoothed data is generated by smoothing the luminance image data. Here, the luminance value of the pixel determined to be included in the first type specific area and the luminance value of the pixel determined to be included in the second type specific area are not smoothed, and the first type specific area is not smoothed. The luminance values of pixels that are not determined to be included in the region and that are not determined to be included in the second type specific region are smoothed. Edge pixels are detected using the smoothed data. [Selection] Figure 17

Description

  The present specification relates to a technique for detecting an edge in an image.

  Conventionally, various image processing has been proposed. For example, there has been proposed a technique for classifying an image into characters and halftone dots by extracting character edges using a differential filter or the like, and accumulating isolated amounts to extract halftone dots for printing. Further, a technique has been proposed in which a character area and a halftone dot area in an image are distinguished from each other and a character in the halftone dot area is also determined.

JP 2007-311836 A

  By the way, the characters in the image can be expressed in various forms. For example, a character may be expressed in a dark color on a light background. On the other hand, characters may be expressed in a light color in a dark background. It has not been easy to detect an edge of a character that can be expressed in various forms.

  The present specification discloses a technique capable of detecting an edge of a character that can be expressed in various forms.

  This specification discloses the following application examples, for example.

Application Example 1 An image processing apparatus, an input image data acquisition unit that acquires input image data representing an input image, and a plurality of brightness levels of a plurality of pixels using the input image data A luminance image data acquisition unit that acquires luminance image data including individual luminance values, and a first determination unit that determines whether or not the target pixel is included in the first type specific region using the luminance image data The first type of specific region is a region where a plurality of pixels are continuous, and the first type of specific region is a region whose brightness is darker than that of the second type of specific region. A second determination unit that determines whether or not the pixel of interest is included in the second type of specific region by using the luminance image data; and a plurality of the second type of specific regions The second type of specific area is the previous area. The second determination unit, which is a region that is brighter than the first type specific region, and a smoothed data generation unit that generates smoothed data by smoothing the luminance image data, Without smoothing the luminance value of the pixel determined to be included in the first type of specific region and the luminance value of the pixel determined to be included in the second type of specific region among the plurality of pixels Generating the smoothed data by smoothing luminance values of pixels that are not determined to be included in the first type of specific area and that are not determined to be included in the second type of specific area; An image processing apparatus comprising: the smoothed data generation unit; and a detection unit that detects edge pixels representing edges in the input image using the smoothed data.

  According to this configuration, the luminance value of the pixel determined to be included in the first type specific area and the luminance value of the pixel determined to be included in the second type specific area are not smoothed. Pixels that represent the edge of a character that is represented by a dark color (that is, a low-intensity color) on a light background, and pixels that represent the edge of a character that is represented by a light color (that is, a high-intensity color) on a dark background Both can be detected appropriately.

  The technology disclosed in the present specification can be realized in various modes. For example, an image processing method and an image processing apparatus, a computer program for realizing the function of the method or the apparatus, and the computer It can be realized in the form of a recording medium on which the program is recorded (for example, a recording medium that is not temporary).

2 is a block diagram illustrating a configuration of the multifunction peripheral 100. FIG. It is a block block diagram of the area | region separation process part 103 in embodiment. 4 is a block diagram illustrating a configuration of a character determination unit 1003. FIG. It is a figure for demonstrating the process of a character determination part. It is a figure which shows the signal of each process of a character determination part. It is explanatory drawing of a table. 3 is a block configuration diagram of a halftone dot determination unit 1004. FIG. It is a figure for demonstrating the process of a halftone dot determination part. It is a 1st figure which shows an example of pattern matching. It is a figure which shows an example of the result of matching with an inner edge signal and the determination pattern 1710. FIG. 10 is an explanatory diagram of processing by an OR processing unit 1207. It is a block diagram which shows the structure of the integrating | accumulating process part 1209 and the threshold value determination part 1211. It is a figure which shows an example of a process with the threshold value determination part 1211 and the comprehensive determination part 1213. It is a figure which shows an example of a process with the threshold value determination part 1211 and the comprehensive determination part 1213. It is a figure which shows an example of a process with the threshold value determination part 1211 and the comprehensive determination part 1213. 10 is a block diagram showing a configuration of a halftone dot character determination unit 1005. FIG. It is a block diagram which shows the structure of the adaptive smoothing process part 2401. FIG. It is explanatory drawing of a line pattern. It is a figure which shows the example of the determination using the line pattern 2600 of FIG. It is a figure which shows the example of the determination using the line patterns 2610-2630 of FIG.18 (B)-(D). It is a figure which shows the example of the determination using the line pattern 2600 of FIG. It is a figure which shows the example of the determination using the line patterns 2610-2630 of FIG.18 (B)-(D). It is explanatory drawing which shows an example of the frequency characteristic 2900 of the smoothing by the adaptive smoothing process part 2401. FIG. It is a figure which shows the example of the halftone character determination by the halftone character determination part 1005. FIG. It is a figure which shows the example of the halftone character determination by the halftone character determination part 1005. FIG. 3 is a block diagram illustrating a configuration of an output processing unit 110. FIG. It is a flowchart which shows another Example of the process which detects the pixel of a 1st type specific area, and the pixel of a 2nd type specific area. It is explanatory drawing which shows the example of a some filter. It is explanatory drawing of the process which determines the initial value Wdiw for bright lines.

A. First embodiment:
A-1. Configuration of MFP 100 Next, an embodiment will be described based on examples. FIG. 1 is a block diagram illustrating a configuration of a multifunction peripheral 100 as an image processing apparatus according to an embodiment.

  The MFP 100 includes a control unit 120, a reading execution unit 130, a printing execution unit 140, a communication interface (communication IF) 150, a display unit such as a liquid crystal display (not shown), and an operation unit including a touch panel and buttons (not shown). And.

  The control unit 120 includes a main processor 101, a plurality of image processing units 102 to 110, a plurality of volatile memories 111 to 116 such as a DRAM, and a nonvolatile memory 117 such as a flash memory.

  The main processor 101 executes the computer program PG stored in the nonvolatile memory 117, thereby controlling the entire multifunction peripheral 100 including the reading execution unit 130, the print execution unit 140, and the image processing units 102 to 110. CPU.

  The plurality of image processing units 102 to 110 are hardware circuits such as an ASIC that operate according to the control of the main processor 101. Volatile memories 111 to 116 are memories used by the main processor 101 and the image processing units 102 to 110. An outline of processing using these image processing units and memories will be described later.

  The non-volatile memory 117 stores the computer program PG described above. The computer program PG is a control program that causes the main processor 101 to control the multifunction peripheral 100. The computer program PG is provided in a form stored in advance in the non-volatile memory 117 when the multifunction device 200 is manufactured. Instead of this, the computer program PG may be provided in a form downloaded from a server, or may be provided in a form stored in a DVD-ROM or the like.

  The reading execution unit 130 generates scan data by optically reading a document using a one-dimensional image sensor under the control of the main processor 101. The scan data includes a plurality of pixel values, and each of the plurality of pixel values represents the color of the pixel by an RGB color system color value (also referred to as an RGB value). That is, the scan data is RGB image data. The RGB value of one pixel includes, for example, three color component values of red (R), green (G), and blue (B) (hereinafter also referred to as R value, G value, and B value). It is out. In this embodiment, the number of gradations of each component value is 256 gradations.

  The print execution unit 140 uses a plurality of types of inks, specifically, cyan (C), magenta (M), yellow (Y), and black (K) inks as color materials according to the control of the main processor 101. Thus, an image is printed by ejecting ink as a color material onto a print medium such as paper (also called an inkjet method). In a modified example, the print execution unit 140 may be a laser type print execution unit that prints an image on a print medium by a laser method using toner as a color material.

  The communication IF 150 is an interface for communicating with an external device such as a user terminal device (not shown), for example, a network interface according to the Ethernet (registered trademark) standard.

A-2. Overview of Processing of Multifunction Device 100 Copy processing executed by the multifunction device 100, that is, processing for generating scan data indicating a document using the reading execution unit 130, and printing an image indicating the document using the scan data The outline of will be described. The document of this embodiment is a printed material on which an image is printed by, for example, the multifunction peripheral 100 or a printer (not shown).

  The user sets a document to be copied on the sheet feeder of the reading execution unit 130, and inputs a copy start instruction via an operation unit (not shown). The reading execution unit 130 generates scan data as target image data by reading one document at a time in response to a start instruction. The generated scan data is output to the first input processing unit 102.

  The first input processing unit 102 performs well-known image processing such as shading correction and color correction on the scan data output from the reading execution unit 130, and sends the image processed scan data to the region separation processing unit 103. Send.

  The region separation processing unit 103 performs region separation processing on the scan data output from the first input processing unit 102. The region separation processing unit 103 detects image features such as a photograph (natural image) region, a character region, a halftone dot region, and a character region within a halftone dot for each pixel of the scanned image indicated by the scan data, and Flag data representing an attribute is generated as a character flag 1007, a halftone dot flag 1008, and a halftone dot character flag 1009 described later. The flag data is output to the second input processing unit 104 and stored in the first flag memory 112.

  The second input processing unit 104 performs appropriate image processing for each region on the scan data output from the first input processing unit 102 based on the flag data output from the region separation processing unit 103, and performs processing. Generate finished image data. The processed image data is stored in the first image memory 111. For example, the second input processing unit 104 performs processing for emphasizing high-frequency components of the image to enhance the sharpness of the character with respect to the character region, and performs low-pass filter processing with respect to the dot region. A so-called moire component included in the scan image is removed by the low-pass filter processing. Switching between these processes is performed on a pixel basis based on the flag data generated by the region separation processing unit 103.

  When processed image data for one page is stored in the first image memory 111, the data compression unit 107 compresses the processed image data using, for example, an irreversible compression method (for example, JPEG method). The compressed processed image data is stored in the main memory 113.

  When one page of flag data is stored in the first flag memory 112, the data compression unit 107 compresses the flag data using a lossless compression method (for example, the ZIP method), and converts the compressed flag data into the first flag memory 112. And stored in the main memory 113.

  By performing the above processing, the processed image data and the compressed flag data of each document are accumulated in the main memory 113. In parallel with this accumulation process, the data decoding unit 109 acquires the processed image data and the compressed flag data from the main memory 113, and performs the decoding process. At this time, the resolution conversion unit 108 performs resolution conversion for converting the resolution of the processed image data on the decoded processed image data as necessary.

  The processed image data whose decoding and resolution have been converted are stored in the second image memory 116, and the decoded flag data is stored in the second flag memory 115.

  The output processing unit 110 performs generation processing including color conversion processing and halftone processing on the processed image data (RGB values) stored in the second image memory 116 to generate print data. In the process of generating print data, the output processing unit 110 refers to the flag data stored in the second flag memory 115, recognizes the attribute of the area to which each pixel belongs, and executes image processing suitable for each area. can do. As a result, appropriate print data is generated. The generated print data is output to the print execution unit 140. As a result, the print execution unit 140 prints an image showing the document on the paper.

  The above is for copy processing, but the multi-function device 100 receives a print job transmitted from a terminal device such as a user terminal device via the communication IF 150, and displays an image based on the print job. It also functions as a so-called network printer for printing.

  In this case, the print job received via the communication IF 150 includes image data in PDL format. A print job including image data in the PDL format is supplied to the interpreter 106. The interpreter 106 interprets PDL format image data included in the received print job, and converts it into a drawing command (intermediate data) that can be interpreted by the RIP 105. The RIP 105 performs drawing processing (rasterization processing) based on this drawing command to generate RGB image data. This RGB image data is stored in the first image memory 111. The RIP 105 generates flag data and stores it in the first flag memory 112. In PDL format print data, characters, photos (natural images), halftone dots, and the like are defined by print commands, so the RIP 105 can easily generate flag data.

A-3. Processing by Region Separation Processing Unit 103 Processing executed by the region separation processing unit 103 will be described. FIG. 2 is a block configuration diagram of the region separation processing unit 103 in the embodiment.

  An input signal 1001 in units of pixels indicating the value of each pixel of the scan data output from the first input processing unit 102 is input to the luminance data acquisition unit 1002 for each pixel. That is, the input signal 1001 is a signal indicating the RGB value of each pixel of the scan data. Based on the RGB values indicated by the input signal 1001, the luminance data acquisition unit 1002 generates a signal indicating a luminance value, which is the degree of brightness of the pixel, as a determination signal for each pixel for attribute determination. The luminance value is acquired using a plurality of component values (R value, G value, B value) included in the RGB value. For example, the G value is used as the luminance value. Instead of this, the luminance value may be calculated by an operation such as (R + 2 × G + B) / 4. Alternatively, the RGB color system color value may be converted into the Lab color system color value, and the obtained L value may be used as the luminance value. It can be said that the luminance data acquisition unit 1002 is a circuit that acquires luminance image data representing the luminance values of a plurality of pixels using scan data input for each pixel.

  The determination signal indicating the luminance value generated by the luminance data acquisition unit 1002 is output to the character determination unit 1003, the halftone dot determination unit 1004, and the halftone dot character determination unit 1005, respectively. The character determination unit 1003, the halftone dot determination unit 1004, and the halftone dot character determination unit 1005 execute character determination, halftone dot determination, and halftone dot character determination, respectively, using the supplied determination signal. The result signal indicating the determination result is output to the attribute flag generation unit 1006.

  The attribute flag generation unit 1006 generates an attribute flag using the result signals from the determination units 1003 to 1005. In the embodiment, a character flag 1007, a halftone dot flag 1008, and a halftone dot character flag 1009 are generated. The character flag 1007 is a flag indicating whether or not the corresponding pixel in the scan data is a character pixel constituting the character area. A halftone dot flag 1008 is a flag indicating whether or not the corresponding pixel in the scan data is a halftone dot pixel constituting a halftone dot region. The halftone dot character flag 1009 is a flag indicating whether or not the corresponding pixel in the scan data is a halftone dot character pixel constituting a halftone dot character region. Based on these attribute flags, it is possible to perform optimum image processing according to the characteristics of the image included in the scanned image. Hereinafter, the character determination unit 1003, the halftone dot determination unit 1004, and the halftone dot character determination unit 1005 will be described in more detail.

A-3-1. Description of Character Determination Unit 1003 FIG. 3 is a block diagram showing a configuration of the character determination unit 1003.

  The edge enhancement unit 1102 receives a determination signal (luminance value) from the luminance data acquisition unit 1002. The edge enhancement unit 1102 performs edge enhancement processing on the determination signal and outputs an edge enhancement signal. In this edge enhancement process, a filter process for enhancing / extracting a desired frequency component of luminance image data is performed. In this embodiment, a secondary differential filter such as a Laplacian filter is used. For this purpose, the edge enhancement unit 1102 includes a buffer memory for storing a plurality of determination signals indicating luminance values of a plurality of pixels of luminance image data necessary for the filtering process.

  The edge enhancement signal output from the edge enhancement unit 1102 is input to threshold determination units 1103 and 1104 that compare a value indicated by the edge enhancement signal with a threshold value. The threshold value determination unit 1103 is set with a positive threshold value, and the threshold value determination unit 1104 is set with a negative value threshold value.

  FIG. 4 is a diagram for explaining the processing of the character determination unit. When a secondary differential filter is used in the edge enhancement processing, the pixel value indicated by the edge enhancement signal may be a positive value or a negative value. In FIG. 4A, an edge boundary portion 1501 represents an edge between a character 1502 (for example, black) and a background 1503 on a background 1503 (for example, white). FIG. 4B shows luminance image data (that is, a determination signal) 1504 corresponding to the cross section 1510 of the edge boundary 1501. In this example, the luminance image data value (luminance value) in the character 1502 portion is a relatively small value 1505, and the luminance image data value in the background 1503 portion is a relatively large value 1506. FIG. 4C shows edge extraction data 1507 (that is, data indicated by an edge enhancement signal) obtained by performing edge enhancement processing on the luminance image data 1504 of FIG. 4B using a secondary differential filter. )It is shown. The edge extraction data 1507 has a positive value 1508 on the character 1502 side and a negative value 1509 on the background 1503 side. The threshold determination unit 1103 outputs an inner edge signal indicating “1” when the value indicated by the edge enhancement signal exceeds the positive threshold, and indicates “0” when the value is equal to or less than the positive threshold. The inner edge signal is output. The inner edge signal indicating “1” indicates that the corresponding pixel is a small value side pixel (inner edge pixel) that constitutes an edge that changes from a relatively small value to a relatively large value in the luminance image data. Indicates. In the example of FIG. 4, the pixel on the character side that forms the edge between the character 1502 and the background 1503 is the inner edge pixel. The threshold determination unit 1104 outputs an outer edge signal indicating “1” when the value indicated by the edge enhancement signal is below the negative threshold, and indicates “0” when the value is equal to or greater than the negative threshold. Outputs an outer edge signal. The outer edge signal indicating “1” is a pixel on the large value side (also referred to as an outer edge pixel) that constitutes an edge in which the corresponding pixel changes from a relatively small value to a relatively large value in the luminance image data. Indicates that there is. In the example of FIG. 4, the pixel on the ground side that forms the edge between the character 1502 and the ground 1503 is the outer edge pixel.

  The inner edge signal output from the threshold determination unit 1103 is output to the area integration unit 1105, and the outer edge signal output from the threshold determination unit 1104 is output to the area integration unit 1106. The area integration unit 1105 outputs, for each pixel, an integration signal obtained by integrating nine inner edge signals corresponding to pixels in the peripheral range of 3 vertical pixels × 3 horizontal pixels centered on the target pixel. The area integration unit 1106 outputs an integration signal obtained by integrating the nine outer edge signals corresponding to the pixels in the peripheral range of the target pixel for each pixel. Therefore, the integration signals output from the area integration units 1105 and 1106 indicate the number of inner edge pixels and the number of outer edge pixels existing in the peripheral range of the corresponding pixel, respectively. Therefore, the value indicated by the integrated signal is in the range of 0 to 9.

  The threshold determination units 1107 and 1108 compare the value indicated by the integration signal output from the area integration units 1105 and 1106 with the threshold value, and output an integration determination signal indicating the result. For example, a value “2” is set as each threshold value. For this reason, the integration determination signal output from the threshold determination unit 1107 has a positive value (for example, “1”) when there are two or more inner edge pixels in the peripheral range of the corresponding pixel. Show. The integration determination signal output from the threshold determination unit 1108 indicates a positive value (for example, “1”) when there are two or more outer edge pixels in the peripheral range of the corresponding pixel.

  A specific example of the process will be described with reference to FIG. FIG. 5 is a diagram illustrating signals of each process of the character determination unit. FIG. 5A shows a part 1301 of the edge of the character on the base that appears in the luminance image indicated by the luminance image data.

  FIG. 5B shows a result of edge enhancement of the determination signal (that is, a signal indicating a luminance value) by the edge enhancement unit 1102 and processing by the threshold determination unit 1104, that is, a result of specifying the outer edge pixel. As illustrated, the region 1302 represents a region outside the character edge (outer edge region).

  FIG. 5C shows the result of processing the outer edge signal from the threshold determination unit 1104 by the area integration unit 1106 and processing the integration signal from the area integration unit 1106 by the threshold determination unit 1108, that is, area integration of the outer edge. The determination result is shown. As shown, region 1304 is the result of extending outer edge region 1302.

  FIG. 5D shows the result of processing the edge of the determination signal (that is, the signal indicating the minimum component value Vmin) by the edge enhancement unit 1102 and processing by the threshold determination unit 1103, that is, the result of specifying the inner edge pixel. Yes. As illustrated, a region 1303 represents a region inside the edge of the character (inner edge region).

  FIG. 5E shows the result of processing the inner edge signal from the threshold determining unit 1103 by the area integrating unit 1105 and processing the integrated signal from the area integrating unit 1105 by the threshold determining unit 1107, that is, area integration of the inner edge. The determination result is shown. As shown in the figure, the area 1305 is the result of expanding the inner edge area 1303.

  Three types of signals output from the threshold determination units 1103, 1107, and 1108, that is, the inner edge signal, the inner edge integration determination signal, and the outer edge integration determination signal are input to the overall determination unit 1109. The The overall determination unit 1109 generates a character determination signal 1110 that is a result signal indicating the determination result of the character determination unit 1003. FIG. 5F shows a region 1306 indicated as a character by the character determination signal.

Based on the inner edge signal, the inner edge integration determination signal, and the outer edge integration determination signal, the overall determination unit 1109 outputs a character determination signal 1110 according to a determined logic. For example, the character determination signal 1110 is a signal indicating that it is a character (“1” in the present embodiment) or a signal indicating that it is not a character (“0” in this embodiment) according to the following logic. It is determined.
(1) The corresponding pixel is the inner edge. → "1"
(2) The corresponding pixel is the outer edge, and there are two or more outer edges in the peripheral area of the vertical 3 pixels × vertical 3 pixels. → "0"
(3) It does not correspond to any of the above (1) and (2), and there are two or more inner edges in the peripheral area of the vertical 3 pixels × vertical 3 pixels. → "1"
(4) Does not correspond to any of the above (1) to (3). → "0"
In order to realize the above, the comprehensive determination unit 1109 holds the table shown in FIG. As shown in the figure, the character determination signal 1110 is determined according to the result of the input inner edge signal, inner edge integration determination signal, and outer edge integration determination signal. By performing as described above, it is possible to suitably extract characters.

A-3-2. Description of Halftone Determination Unit 1004 FIG. 7 is a block configuration diagram of the halftone dot determination unit 1004 in the embodiment.

  The edge enhancement unit 1202 receives a determination signal (luminance value) from the luminance data acquisition unit 1002. The edge enhancement unit 1202 performs edge enhancement processing on the determination signal and outputs an edge enhancement signal. The edge enhancement processing by the edge enhancement unit 1202 is the same processing as the edge enhancement processing by the edge enhancement unit 1102 in FIG. 3, for example, filter processing using a secondary differential filter such as a Laplacian filter.

  The edge enhancement signal output from the edge enhancement unit 1202 is input to threshold determination units 1203 and 1204 that compare a value indicated by the edge enhancement signal with a threshold value. The threshold value determination unit 1203 is set with a positive threshold value, and the threshold value determination unit 1204 is set with a negative value threshold value.

  FIG. 8 is a diagram for explaining the processing of the halftone dot determination unit. When the secondary differential filter is used in the edge enhancement process, as described above, the pixel value indicated by the edge enhancement signal may be a positive value or a negative value. In FIG. 8A, an edge boundary portion 1601 represents the edge of one halftone dot 1602 on the background 1603 (for example, white) and the background 1603. FIG. 8B shows luminance image data (that is, a determination signal) 1604 for the cross section 1610 of the halftone edge boundary 1601. In this example, the luminance image data value (luminance value) in the halftone dot 1602 portion is a relatively small value 1605, and the luminance image data value in the background 1603 portion is a relatively large value 1606. FIG. 8C shows edge extraction data 1607 obtained by performing edge enhancement processing on the luminance image data 1604 of FIG. 8B using a secondary differential filter. The edge extraction data 1607 has a positive value 1608 on the halftone dot 1602 side and a negative value 1609 on the background 1603 side. The threshold determination unit 1203 outputs an inner edge signal in the same manner as the threshold determination unit 1103 in FIG. The threshold determination unit 1204 outputs an outer edge signal, similar to the threshold determination unit 1104 in FIG. In the example of FIG. 6, the pixel on the halftone dot side that forms the edge between the halftone dot 1602 and the background 1603 is the inner edge pixel. Further, the pixel on the ground side that forms the edge between the halftone dot 1602 and the ground layer 1603 is an outer edge pixel.

  The inner edge signal output from the threshold determination unit 1203 is input to the isolation amount determination unit 1205. The outer edge signal output from the threshold determination unit 1204 is input to the isolation amount determination unit 1206.

  The isolation amount determination unit 1205 performs pattern matching processing on the inner edge signal from the threshold determination unit 1203. Since halftone dots included in a document range from those having a relatively small number of lines per unit area to those having a relatively large number of lines per unit area, the sizes and intervals of the halftone dots vary depending on the document. Therefore, pattern matching is performed with a plurality of patterns so that halftone dots with any number of lines can be detected. For halftone dots with a small number of lines, matching is performed with a relatively large pattern to determine whether the dot is a halftone dot. For halftone dots with multiple lines, matching is performed with a relatively small pattern to determine whether the dot is a halftone dot. Further, since the shape of the halftone dot changes depending on the density to be expressed on the document, a plurality of determination levels are used for matching so as to cope with it.

  FIG. 9 is a first diagram illustrating an example of pattern matching. FIG. 9A is an example of the result of the threshold determination unit 1203. FIG. 9A shows an inner edge signal 1700 within the range in order to explain an example of pattern matching in a range of 4 vertical pixels × 4 horizontal pixels. Among them, the pixels (inner edge pixels) from which the inner edge signal indicating “1” is output by the threshold determination unit 1203 are the four pixels in the hatched region 1702. A pixel (a non-inner edge pixel) from which an inner edge signal indicating “0” is output is a pixel in a region 1703 that is not hatched. A target pixel to be matched is a pixel 1701.

  Next, an example of a determination pattern of the same size is shown in FIG. In the illustrated determination pattern 1710, four black pixels 1712 are pixels that should be inner edge pixels in the pattern. Eight white pixels 1713 are pixels that should be non-inner edge pixels. The pixels at the four corners are arbitrary pixels 1714 that may be inner edge pixels or non-inner edge pixels. Using this as a basic pattern, each of the black pixel 1712 and the white pixel 1713 has a plurality of matching levels, and the determination level is adjusted.

  FIG. 10 is a diagram illustrating an example of the inner edge signal, the determination pattern 1710 in FIG. 9B, and the matching result.

  When the black pixel 1712 in the determination pattern 1710 and the inner edge pixel in the inner edge signal match at least N1 pixels (N1 is an integer of 1 to 3) of the four pixels, the black pixel determination is valid. . When the white pixel 1713 in the determination pattern 1710 and the non-edge pixel in the inner edge signal match N2 pixels or more (N2 is an integer of 1 to 7) among the 8 pixels, the white pixel determination is valid. Only when both the black pixel determination and the white pixel determination are valid, the isolation amount determination signal indicating the determination result of the target pixel 1801 is set to “1”. The determination level can be arbitrarily adjusted by adjusting the values of N1 and N2. In this embodiment, N1 = 3 and N2 = 6.

  In the case of the inner edge signal 1810 in FIG. 10A, the black pixel determination is valid (number of matches = 4) and the white pixel determination is valid (number of matches = 8). 1 ".

  In the case of the inner edge signal 1820 in FIG. 10B, the black pixel determination is valid (number of matches = 4) and the white pixel determination is valid (number of matches = 7). 1 ".

  In the case of the inner edge signal 1830 in FIG. 10C, the black pixel determination is valid (number of matches = 3) and the white pixel determination is valid (number of matches = 8). 1 ".

  In the case of the inner edge signal 1840 in FIG. 10D, since the black pixel determination is invalid (match number = 2) and the white pixel determination is valid (match number = 8), the isolation amount determination signal is “ 0 ".

  Such pattern matching is performed using determination patterns of a plurality of sizes. In the embodiment, the inner edge signal has been described, but the same pattern matching is performed for the outer edge signal. At this time, the adjustment value of the determination pattern and the determination level can be arbitrarily set.

  The isolation amount determination signals output from the isolation amount determination units 1205 and 1206 are input to the OR processing unit 1207 in FIG. The OR processing unit 1207 calculates the logical sum of the isolation amount determination signals within the range of 3 vertical pixels × 3 horizontal pixels, and determines whether there is a pixel with the isolation amount determination signal “1” within the range. . As a result of the determination, the OR processing unit 1207 sets the OR processing signal of the target pixel to be determined to “1” if there is one or more pixels whose isolation amount determination signal is “1” within the range, and the isolation amount If there is no pixel having the determination signal “1” within the range, the OR processing signal of the target pixel to be determined is set to “0”.

  FIG. 11 is an explanatory diagram of processing by the OR processing unit 1207. In the example of FIG. 11A, only the isolated amount determination signal of the center pixel of interest in the range of 3 vertical pixels × 3 horizontal pixels is “1”. Therefore, the output OR processing signal is “1”. In the example of FIG. 11B, the isolation amount determination signal of the pixel of interest at the center is “0”, but the isolation amount determination signal of the upper right pixel within the range of 3 vertical pixels × 3 horizontal pixels is “1”. It is. Therefore, the output OR processing signal is “1”. In the example of FIG. 11C, the isolation amount determination signal of the center pixel of interest is “0”, but the isolation amount determination signals of the eight surrounding pixels are “1”. Therefore, the output OR processing signal is “1”.

  The OR processing signal output from the OR processing unit 1207 is input to the integration processing unit 1209. FIG. 12 is a block diagram illustrating configurations of the integration processing unit 1209 and the threshold determination unit 1211.

  The integration processing unit 1209 outputs, for each pixel, an integration signal obtained by integrating a plurality of OR processing signals corresponding to pixels within an integration range of a plurality of types with the pixel of interest as the center. The signal of the OR processing signal is integrated using a plurality of integration ranges. For example, the first integration unit 2011 uses a range of 9 vertical pixels × 9 horizontal pixels as the integration range. Therefore, the integration determination signal output from the first integration unit 2011 is in the range of 0 to 81. The second integration unit 2012 and the third integration unit 2013 use a range of 15 pixels vertical by 15 pixels horizontal and a range of 21 pixels vertical by 21 pixels horizontal as the integration ranges, respectively. Therefore, the integration signals output from the second integration unit 2012 and the third integration unit 2013 are in the range of 0 to 225 and 0 to 441, respectively. Integration signals output from the first integration unit 2011, the second integration unit 2012, and the third integration unit 2013 are respectively sent to the first determination unit 2021, the second determination unit 2022, and the third determination unit 2023 of the threshold determination unit 1211. Is entered.

  The first determination unit 2021, the second determination unit 2022, and the third determination unit 2023 compare the value indicated by the integrated signal input thereto with the threshold value, and output an integrated determination signal indicating the result. I do. Specifically, the integration determination signal output from the first determination unit 2021, the second determination unit 2022, and the third determination unit 2023 indicates that the value indicated by the integration signal is equal to or greater than a set threshold value. “1”, and when it is less than the threshold, it is “0”. The threshold values used in the first determination unit 2021, the second determination unit 2022, and the third determination unit 2023 are different from each other, and are set to “5”, “12”, and “23”, respectively, in this embodiment. ing. Thus, each determination unit 2021, 2022, 2023 outputs an integration determination signal indicating “1” when the value indicated by the integration signal exceeds about 5% of the number of pixels within the integration range, and the integration signal is When the indicated value is about 5% or less of the number of pixels in the integration range, an integration determination signal indicating “0” is output.

  The integration determination signals output from the first determination unit 2021, the second determination unit 2022, and the third determination unit 2023 are input to the comprehensive determination unit 1213 (FIGS. 7 and 12). The overall determination unit 1213 generates a halftone dot determination signal 1215 that is a result signal indicating the determination result of the halftone dot determination unit 1004 based on the three input integration determination signals. The halftone dot determination signal 1215 is either a value “1” indicating that the target pixel is a halftone pixel, or a value “0” indicating that the target pixel is not a halftone pixel.

  13 to 15 are diagrams illustrating an example of processing by the threshold determination unit 1211 and the comprehensive determination unit 1213. FIG. 13A conceptually shows an OR processing signal 2100 output by the OR processing unit 1207. This OR processing signal 2100 is an example when a region including a halftone dot having a relatively low density is processed. Cross-hatched pixels indicate pixels whose corresponding OR processing signal is “1”, and pixels that are not cross-hatched indicate pixels whose corresponding OR processing signal is “0”. In the figure, an integration range 2102 of 9 vertical pixels × 9 horizontal pixels, an integration range 2103 of 15 vertical pixels × 15 horizontal pixels, and an integration range 2104 of 21 vertical pixels × 21 horizontal pixels, centered on the target pixel 2101, , Are shown respectively. FIG. 13B shows the integration results in the respective integration ranges, that is, the values of the integration signals output from the first integration unit 2011, the second integration unit 2012, and the third integration unit 2013. The integration result of the first integration unit 2011 (9 × 9) is 12 pixels, the integration result of the second integration unit 2012 (15 × 15) is 40 pixels, and the integration result of the third integration unit 2013 is 76 pixels.

  Therefore, as shown in FIG. 13B, since the integration result “12” of the first integration unit 2011 (9 × 9) is equal to or greater than the threshold value “5”, the first determination unit 2021 (9 × 9) The integration determination signal to be output is “1” (halftone dot). Similarly, the integration result “40” of the second integration unit 2012 (15 × 15) is also equal to or greater than the threshold “12”, and the integration result “76” of the third integration unit 2013 (21 × 21) is also the threshold “23”. Therefore, the integrated determination signals output from the corresponding second determination unit 2022 and third determination unit 2023 are both “1”.

  Based on the above three types of results, the overall determination unit 1213 determines whether or not the target pixel is finally a halftone dot, and generates a halftone dot determination signal 1215 indicating the determination result. In this embodiment, the comprehensive determination unit 1213 creates two pairs from three types of results, as shown in FIG. For example, a first pair is created by the integration determination signal of the first determination unit 2021 and the second determination unit 2022, and a second pair is generated by the integration determination signal of the second determination unit 2022 and the third determination unit 2023. The comprehensive determination unit 1213 calculates the logical product (AND) of the integration determination signals of each set. The logical product of the first pair is “1” (halftone dot), and the logical product of the second pair is also “1” (halftone dot). The overall determination unit 1213 further generates a logical sum of the logical product of the first pair and the logical product of the second pair as a halftone determination signal 1215. In the example of FIG. 13, since both logical products are “1”, the halftone dot determination signal 1215 of the target pixel is “1” (halftone dot).

  FIG. 14A conceptually shows an OR processing signal 2200 corresponding to a region including a halftone dot having a density similar to that in FIG. 13 and including an edge 2205 of a character. As shown in the figure, when there is a character edge 2205, an isolated dot does not occur near the edge, so the OR processing signal near the character edge is “0”. In this case, as shown in FIG. 14B, the integration result of the first integration unit 2011 (9 × 9) is 4 pixels, and the integration result of the second integration unit 2012 (15 × 15) is The number of pixels is 32, and the integration result of the third integration unit 2013 is 60 pixels.

  Therefore, as shown in FIG. 14B, since the integration result “4” of the first integration unit 2011 (9 × 9) is less than the threshold value “5”, the first determination unit 2021 (9 × 9) The integration determination signal to be output is “0” (non-halftone dot). The integration result “32” of the second integration unit 2012 (15 × 15) is equal to or greater than the threshold “12”, and the integration result “60” of the third integration unit 2013 (21 × 21) is also equal to or greater than the threshold “23”. Therefore, the integration determination signals output from the corresponding second determination unit 2022 and third determination unit 2023 are both “1” (halftone dot).

  As shown in FIG. 14C, the logical product of the first pair is “0” (non-halftone dot), and the logical product of the second pair is “1” (halftone dot). The logical sum of the logical product of the first pair and the logical product of the second pair, that is, the halftone dot determination signal 1215 of the target pixel is “1” (halftone dot). As described above, since the edge of the character is included, even if the integration result of one integration range indicates a non-halftone dot, it is finally determined as a halftone dot. In this way, by using a plurality of integration ranges and a plurality of threshold values, it can be appropriately detected as a halftone dot.

  FIG. 15A conceptually illustrates an OR processing signal 2300 when the pixel of interest is outside the edge of a region including a halftone dot having a density similar to that in FIG. In this case, as shown in FIG. 15B, the integration result of the first integration unit 2011 (9 × 9) is 0 pixel, and the integration result of the second integration unit 2012 (15 × 15) is There are 12 pixels, and the integration result of the third integration unit 2013 is 20 pixels.

  Therefore, as shown in FIG. 15B, since the integration result “0” of the first integration unit 2011 (9 × 9) is less than the threshold value “5”, the first determination unit 2021 (9 × 9) The integration determination signal to be output is “0” (non-halftone dot). Since the integration result “12” of the second integration unit 2012 (15 × 15) is equal to or greater than the threshold value “12”, the integration determination signal output by the second determination unit 2022 (15 × 15) is “1” (network Point). Since the integration result “20” of the third integration unit 2013 (21 × 21) is less than the threshold “23”, the integration determination signal output by the third determination unit 2023 (21 × 21) is “0” (non- Halftone dot).

  As shown in FIG. 15C, the logical product of the first pair is “0” (non-halftone dot), and the logical product of the second pair is also “0” (non-halftone dot). The logical sum of the logical product of the first pair and the logical product of the second pair, that is, the halftone dot determination signal 1215 of the target pixel is “0” (non-halftone dot). As shown in the example of FIG. 14, when a character edge is included in the halftone dot area included in the accumulation range, the halftone dot at the position overlapping the character edge is removed by the character edge. The total number of Here, when the integration range is expanded, the total number of halftone dots included in the integration range increases, and therefore the ratio of the number of halftone dots removed by the edge of the character to the total number of halftone dots decreases. As a result, the pixels in the halftone area can be appropriately determined as halftone dots. However, when the integration range is expanded, pixels far away from the edge of the area including the halftone dots can also be determined to be halftone dots, so that the area determined to be a halftone dot can be expanded. In this embodiment, by using a plurality of integration ranges and a plurality of threshold values, pixels outside the edge of the halftone dot can be appropriately detected as non-halftone dots.

  The combination of integration processes described here is merely an example, and the present invention is not limited to this. Combinations can be freely configured according to the purpose. In the description, the results of three types of integration processing are used, so two sets of logical products are used. However, the present invention is not limited to this. The number of inputs and their logical operation units can be freely configured. Furthermore, the combination of logical product and logical ring is only an example, and the present invention is not limited to this. You can freely combine logical products and logical sums.

A-3-2. Description of Intradot Character Determination Unit 1005 FIG. 16 is a block diagram showing the configuration of the in-halftone character determination unit 1005.

  The adaptive smoothing processing unit 2401 receives the determination signal from the luminance data acquisition unit 1002 and performs an adaptive smoothing process. Here, digital filter processing for smoothing a desired frequency component of scan data while adaptively excluding characters / thin lines is performed.

  The smoothing signal output from the adaptive smoothing processing unit 2401 is input to the edge enhancement unit 2402. The edge enhancement unit 2402 performs edge enhancement processing on the smoothing signal and outputs an edge enhancement signal. The edge enhancement processing by the edge enhancement unit 2402 is the same processing as the edge enhancement processing by the edge enhancement unit 1102 in FIG. 3, for example, filter processing using a secondary differential filter such as a Laplacian filter.

  The edge enhancement signal output from the edge enhancement unit 2402 is input to the threshold determination unit 2403. The threshold value determination unit 2403 is set with a positive threshold value.

  As described with reference to FIG. 4, when a secondary differential filter is used in the edge enhancement process, the pixel value indicated by the edge enhancement signal may be a positive value or a negative value. is there. As shown in FIG. 4B, when the character is expressed in a dark color and the background is expressed in a bright color, edge extraction data 1507 ( In other words, the data indicated by the edge emphasis signal has a positive value 1508 on the character 1502 side and a negative value 1509 on the background 1503 side. On the contrary, when the character is expressed in a light color and the background is expressed in a dark color, the edge extraction data becomes a negative value corresponding to the outer edge on the character side and corresponds to the inner edge on the background side. It becomes a positive value (not shown).

  The character determination in the halftone dot is not intended to distinguish whether the edge is on the character side or the background side, but is intended to extract the character itself in the halftone dot region. In this case, it is only necessary to detect one of a positive edge (that is, an inner edge) and a negative edge (that is, an outer edge). In this embodiment, the inner edge is detected. Specifically, the threshold determination unit 2403 outputs an inner edge signal indicating “1” when the value indicated by the edge enhancement signal exceeds the positive threshold, and when the value is equal to or less than the positive threshold, An inner edge signal indicating “0” is output. However, the threshold determination unit 2403 may detect a negative edge (outer edge) without detecting a positive edge (inner edge), and may detect a positive edge (inner edge) and a negative edge (outer edge). Both of them may be detected. Hereinafter, the above process will be described in more detail.

  FIG. 17 is a block diagram illustrating a configuration of the adaptive smoothing processing unit 2401. The determination signal 2501 (luminance value) from the luminance data acquisition unit 1002 is branched into four, and is input to the smoothing processing unit 2502, the selector 2504, the dark line determination unit 2503a, and the bright line determination unit 2503b, respectively. The

  The smoothing processing unit 2502 determines the values of a plurality of pixels within a filter range of vertical M pixels × horizontal M pixels (M is an integer of 2 or more) centered on the target pixel to be processed with respect to the determination signal 2501. The smoothing process is executed using The smoothing process is a process aimed at reducing the sensitivity of a specific frequency band. For example, the smoothing process may be a process using a Gaussian filter or a process using an average value filter. The processing unit 2502 supplies a smoothed signal indicating the result of the smoothing process to the selector 2504.

  The determination signal 2501 input to the selector 2504 is delayed by passing through a delay memory (not shown) in order to correspond to an image delay corresponding to the number of processing lines of processing by the smoothing processing unit 2502, and then the selector 2504. Is input.

  The dark line determination unit 2503a uses a processing buffer for M lines of processing by the smoothing processing unit 2502, and uses vertical N pixels × horizontal L pixels (L and N are integers of 2 or more, and N ≦ M is preferable). In the processing area of (), it is determined whether or not the pixel of interest is a pixel on a vertical line, horizontal line, or diagonal line represented by a dark color. The dark line determination unit 2503a outputs a dark line switching signal to the selector 2504 in accordance with the determination result for these dark lines. When the dark line determination unit 2503a determines that the target pixel is a pixel on these dark lines (also referred to as a dark line pixel), the dark line determination unit 2503a outputs a dark line switching signal indicating “1”, and the target pixel is not a dark line pixel. If determined, a dark line switching signal indicating “0” is output.

  The bright line determination unit 2503b uses a processing buffer for M lines of processing by the smoothing processing unit 2502, and uses vertical N pixels × horizontal L pixels (L and N are integers of 2 or more, and N ≦ M In the (preferred) processing region, it is determined whether or not the pixel of interest is a pixel on a vertical line, horizontal line, or diagonal line expressed in a bright color. The bright line determination unit 2503b outputs a bright line switching signal to the selector 2504 in accordance with the determination result for these bright lines. When the bright line determination unit 2503b determines that the pixel of interest is a pixel on these bright lines (also referred to as a bright line pixel), the bright line determination unit 2503b outputs a bright line switching signal indicating “1”, and the pixel of interest is bright. When it is determined that the pixel is not a line pixel, a bright line switching signal indicating “0” is output.

  The selector 2504 is not filtered with the smoothed signal output from the smoothing processing unit 2502 based on the dark line switching signal and the bright line switching signal output from the dark line determination unit 2503a and the bright line determination unit 2503b, respectively. One of the determination signals 2501 is output. When at least one of the dark line switching signal and the bright line switching signal is “1”, that is, when the dark line determination unit 2503a determines that the pixel of interest is a dark line pixel, the selector 2504 In each of the cases where it is determined in 2503b that the pixel of interest is a bright line pixel, a determination signal 2501 that is not filtered is output. When both the dark line switching signal and the bright line switching signal are “0”, that is, when it is determined that the target pixel is neither a dark line pixel nor a bright line pixel, the selector 2504 outputs a smoothing signal. Output.

  The dark line determination unit 2503a will be described. The dark line determination unit 2503a determines whether or not the target pixel is a dark line pixel by using a line pattern as illustrated in FIGS. The size of the line pattern is the size of the above-mentioned vertical N pixels × horizontal L pixels. In the example of FIG. 18, N = 5 and L = 7. The number N of pixels in the vertical direction (sub-scanning direction) is preferably smaller than the number M of pixels in the sub-scanning direction of the filter range obtained by the smoothing processing unit 2502 from the viewpoint of sharing the memory, but is not limited thereto. The number L of pixels in the horizontal direction (main scanning direction) may be larger than M.

  A line pattern 2600 in FIG. 18A indicates a line pattern for determining a vertical line, and a line pattern 2610 in FIG. 18B indicates a line pattern for determining a horizontal line. Line patterns 2620 and 2630 in FIGS. 18C and 18D represent line patterns for determining diagonal lines that extend in different directions. Here, a determination method for a vertical line will be described using the line pattern 2600. Pixels 2601, 261, 261, 2631 indicate the target pixel.

  The pattern 2600 has a size of 5 vertical pixels × 7 horizontal pixels. The line pattern 2600 includes blocks 2602, 2603, and 2604 each having 5 vertical pixels × 1 horizontal pixel. The target pixel 2601 is arranged at the center of the central block 2603 among the blocks 2602, 2603, and 2604 arranged in the horizontal direction. The dark line determination unit 2503a calculates the sum (ALL), maximum value (MAX), and minimum value (MIN) of the values indicated by the determination signal 2501 for each block. That is, for each block, the sum, the maximum value, and the minimum value of the luminance values of the pixels in the block are calculated.

The dark line determination unit 2503a evaluates the following conditional expressions.
(1) MAX (2602) −MIN (2602) ≦ Threshold 1
(2) MAX (2603) −MIN (2603) ≦ Threshold 1
(3) ALL (2603) ≦ ALL (2602) −threshold value 2
Condition (1) evaluates that there is no difference in pixel values for the five pixels in the block 2602. Condition (2) evaluates that there is no difference in pixel values for the five pixels in the block 2603. Condition (3) compares the sum of the pixel values in the block 2602 with the sum of the pixel values in the pixel block 2603, and the pixel block 2603 has a pixel value relative to the pixel block 2602. To be low. When all the conditions (1) to (3) are satisfied, it is determined that the pixel block 2603 forms a dark vertical line. Conditions (1) to (3) are satisfied when the area of the pixel block 2603 is an area where a plurality of pixels with low average luminance are continuous.

The dark line determination unit 2503a evaluates the following conditional expressions.
(4) MAX (2604) −MIN (2604) ≦ Threshold 1
(5) MAX (2603) −MIN (2603) ≦ Threshold 1
(6) ALL (2603) ≦ ALL (2604) —Threshold 2
Condition (4) evaluates that there is no difference in pixel values for the five pixels in block 2604. Condition (5) evaluates that there is no difference in pixel values for the five pixels in the block 2603. Condition (6) compares the sum of the pixel values in the block 2604 with the sum of the pixel values in the pixel block 2603, and the pixel block 2603 has a pixel value relative to the pixel block 2604. To be low. When all the conditions (4) to (6) are satisfied, it is determined that the pixel block 2603 forms a dark vertical line. Conditions (4) to (6) are satisfied when the area of the pixel block 2603 is an area in which a plurality of pixels with low average luminance are continuous.

  As a result of evaluating the conditions (1) to (3) or the conditions (4) to (6), if it is determined that the block 2603 constitutes a dark vertical line, the dark line determination unit 2503a determines that the pixel of interest 2601 is a dark line pixel located on a dark vertical line.

  FIG. 19 is a diagram illustrating an example of determination using the line pattern 2600 of FIG. 19A to 19C show determination signals (luminance values) 2700, 2710, and 2720 in a vertical 5 pixel × horizontal 7 pixel region corresponding to the line pattern 2600. The determination signal 2700 indicates an image of a portion having a dark vertical line on a white background. A plurality of hatched pixels form a vertical line. For convenience of explanation, when the determination signal indicates a value of 256 gradations of 0 to 255, the white pixel has a corresponding determination signal value of “255”, and the single-hatched pixel has a corresponding determination value. It is assumed that the value of the signal is “128” and the value of the corresponding determination signal is “0” for the cross-hatched pixel. In addition, the threshold value 1 of the above conditions (1) to (6) is 20 and the threshold value 2 is 200.

The result of determining authenticity by applying the above conditions (1) to (6) to the determination signal 2700 is as follows.
(1) MAX (2702) −MIN (2702) ≦ Threshold 1
128−128 ≦ 20 → true (2) MAX (2703) −MIN (2703) ≦ Threshold 1
0-0 ≦ 20 → true (3) ALL (2703) ≦ ALL (2702) −threshold 2
0 × 5 ≦ 128 × 5−200 → true (4) MAX (2704) −MIN (2704) ≦ threshold 1
128−128 ≦ 20 → true (5) MAX (2703) −MIN (2703) ≦ Threshold 1
0-0 ≦ 20 → true (6) ALL (2703) ≦ ALL (2704) −threshold 2
0 × 5 ≦ 128 × 5−200 → true In this way, since all of the conditions (1) to (3) are true, it is determined that the pixel of interest is a dark line pixel located on a dark vertical line. In addition, since all of the conditions (4) to (6) are also true, the pixel of interest is determined to be a dark line pixel located on a dark vertical line from this condition.

A determination signal 2710 in FIG. 19B indicates an image of a portion where a dark horizontal line is present on a white background. The result of determining true / false by applying the above conditions (1) to (6) to the determination signal 2710 is as follows. (1) False, (2) False, (3) False, (4) False, (5) False, (6) False As a result, it is determined that the target pixel is not a dark line pixel located on a dark vertical line. The

The determination signal 2720 in FIG. 19C indicates an image of a portion where a dark diagonal line is present on a white background. The result of determining true / false by applying the above conditions (1) to (6) to the determination signal 2720 is as follows.
(1) False, (2) False, (3) True, (4) False, (5) False, (6) True As a result, it is determined that the pixel of interest is not a dark line pixel located on a dark vertical line. The

  FIG. 20 is a diagram illustrating an example of determination using the line patterns 2610 to 2630 in FIGS. A determination signal 2810 in FIG. 20A indicates an image of a portion having a horizontal line on a white background. The determination signals 2820 and 2830 in FIGS. 20B and 20C show images of a portion having a diagonal line on a white background. Also in the line patterns 2610 to 2630, the pixel of interest 2611, 2621, and 2631 is arranged at the center of the central blocks 2613, 2623, and 2633 among the three blocks arranged in the direction perpendicular to the line extending direction. . In the examples of FIGS. 20A to 20C, although details are omitted, it is determined that all the target pixels are dark line pixels.

  Next, the bright line determination unit 2503b will be described. Similarly to the dark line determination unit 2503a, the bright line determination unit 2503b determines whether or not the target pixel is a bright line pixel using the line patterns illustrated in FIGS. Hereinafter, the determination method about a vertical line is demonstrated using the line pattern 2600. FIG.

  The bright line determination unit 2503b calculates the sum (ALL), maximum value (MAX), and minimum value (MIN) of the values indicated by the determination signal 2501 for each of the blocks 2602, 2603, and 2604 of the line pattern 2600. That is, for each block, the sum, the maximum value, and the minimum value of the luminance values of the pixels in the block are calculated.

The bright line determination unit 2503b evaluates the following conditional expressions.
(1) MAX (2602) −MIN (2602) ≦ Threshold 1
(2) MAX (2603) −MIN (2603) ≦ Threshold 1
(7) ALL (2603) ≧ ALL (2602) + threshold 2
Condition (1) evaluates that there is no difference in pixel values for the five pixels in the block 2602. Condition (2) evaluates that there is no difference in pixel values for the five pixels in the block 2603. Condition (7) compares the sum of the pixel values in the block 2602 with the sum of the pixel values in the pixel block 2603, and the pixel block 2603 has a pixel value relative to the pixel block 2602. Highly rated. When the conditions (1), (2), and (7) are all satisfied, it is determined that the pixel block 2603 forms a bright vertical line. Conditions (1), (2), and (7) are satisfied when the area of the pixel block 2603 is an area in which a plurality of pixels with high average luminance are continuous.

The bright line determination unit 2503b evaluates the following conditional expressions.
(4) MAX (2604) −MIN (2604) ≦ Threshold 1
(5) MAX (2603) −MIN (2603) ≦ Threshold 1
(8) ALL (2603) ≧ ALL (2604) + threshold 2
Condition (4) evaluates that there is no difference in pixel values for the five pixels in block 2604. Condition (5) evaluates that there is no difference in pixel values for the five pixels in the block 2603. Condition (8) compares the sum of the pixel values in the block 2604 with the sum of the pixel values in the pixel block 2603, and the pixel block 2603 has a pixel value relative to the pixel block 2604. Highly rated. When the conditions (4), (5), and (8) are all satisfied, it is determined that the pixel block 2603 forms a bright vertical line. When the area of the pixel block 2603 is an area where a plurality of pixels having high average luminance are continuous, the conditions (4), (5), and (8) are satisfied.

  As a result of evaluating the conditions (1), (2), (7) or the conditions (4), (5), and (8), if it is determined that the block 2603 constitutes a bright vertical line, The bright line determination unit 2503b determines that the target pixel is a bright line pixel located on a bright vertical line.

  FIG. 21 is a diagram illustrating an example of determination using the line pattern 2600 of FIG. 21A to 21C show determination signals (luminance values) 2700w, 2710w, and 2720w in a vertical 5 pixel × horizontal 7 pixel region corresponding to the line pattern 2600. The determination signal 2700w indicates an image of a portion having a bright vertical line on a black background. A plurality of pixels without hatching constitutes a vertical line. As in the example of FIG. 19, the determination signal indicates 256 gradation values from 0 to 255, and the white pixel has a corresponding determination signal value of “255”, and the single-hatched pixel corresponds to Assume that the value of the determination signal is “128”, and the value of the corresponding determination signal is “0” for the cross-hatched pixel. Further, it is assumed that the threshold value 1 is 20 and the threshold value 2 is 200 in the conditions (1), (2), (7), (4), (5), and (8).

The result of determining authenticity by applying the above conditions (1), (2), (7), (4), (5), and (8) to the determination signal 2700 is as follows.
(1) MAX (2702w) −MIN (2702w) ≦ Threshold 1
128−128 ≦ 20 → true (2) MAX (2703w) −MIN (2703w) ≦ threshold 1
255-255 ≦ 20 → true (7) ALL (2703 w) ≧ ALL (2702 w) + threshold 2
255 × 5 ≧ 128 × 5 + 200 → true (4) MAX (2704w) −MIN (2704w) ≦ threshold value 1
128−128 ≦ 20 → true (5) MAX (2703w) −MIN (2703w) ≦ threshold 1
255-255 ≦ 20 → true (8) ALL (2703w) ≧ ALL (2704w) + threshold 2
255 × 5 ≧ 128 × 5 + 200 → true In this way, since all of the conditions (1), (2), and (7) are true, it is determined that the target pixel is a bright line pixel located on a bright vertical line. Is done. Since all of the conditions (4), (5), and (8) are also true, the pixel of interest is determined to be a bright line pixel located on a bright vertical line from this condition.

A determination signal 2710w in FIG. 21B indicates an image of a portion where a bright horizontal line is present on a black background. The result of determining authenticity by applying the above conditions (1), (2), (7), (4), (5), and (8) to the determination signal 2710w is as follows. (1) False, (2) False, (7) False, (4) False, (5) False, (8) False As a result, it is determined that the target pixel is not a bright line pixel located on a bright vertical line. Is done.

A determination signal 2720w in FIG. 21C indicates an image of a portion where a bright diagonal line is present on a black background. The result of determining authenticity by applying the above conditions (1), (2), (7), (4), (5), and (8) to the determination signal 2720w is as follows.
(1) False, (2) False, (7) True, (4) False, (5) False, (8) True As a result, it is determined that the target pixel is not a bright line pixel located on a bright vertical line. Is done.

  FIG. 22 is a diagram illustrating an example of determination using the line patterns 2610 to 2630 in FIGS. A determination signal 2810w in FIG. 22A shows an image of a portion where a bright horizontal line is present on a black background. The determination signals 2820w and 2830w shown in FIGS. 22B and 22C indicate images of a portion having a bright diagonal line on a black background. In the example of FIGS. 22A to 22C, although details are omitted, it is determined that all the target pixels are bright line pixels.

  FIG. 23 shows an example of the frequency characteristic 2900 of smoothing by the adaptive smoothing processing unit 2401. The horizontal axis shows the spatial frequency characteristics, and the vertical axis shows the corresponding frequency response. The higher the frequency band, the smaller the frequency response, and there is no response above a certain spatial frequency (2902). An example of performing halftone dot character determination including processing by the adaptive smoothing processing unit 2401 having such smoothing frequency specification will be described with reference to FIG.

  FIG. 24 is a diagram illustrating an example of halftone dot character determination by the halftone dot character determination unit 1005. An image 3000 in FIG. 24A is obtained by extracting a part of a halftone image. The number of output lines of this halftone dot is high. In terms of the frequency characteristic of the filter shown in FIG. 23, this is a halftone dot having such a characteristic that the spatial frequency is at the position of 2903. Therefore, when the halftone image 3000 is filtered using the filter having the spatial frequency characteristics shown in FIG. 23, the periodic structure of the halftone dot disappears and a smoothed image 3010 is obtained.

  A case where there are dark characters within the same number of lines will be described with reference to FIG. A dark color (for example, black) character is printed in the halftone image 3020, and when the processing by the adaptive smoothing processing unit 2401 is performed on this image, a smoothed image 3030 is obtained. The halftone dot portion is smoothed and the periodic structure of the halftone dot disappears. The character portion is smoothed by the adaptive processing, and the character region 3031 remains clearly. When this image is edge-enhanced by the edge emphasizing unit 2402, an image 3040 shown in FIG. 24C is obtained. In the image 3040, the edge of the character (specifically, the edge on the character side) is emphasized. When an edge-enhanced image is subjected to threshold determination processing by the threshold determination unit 2403, a halftone dot character signal 3050 is obtained. A black part is an area determined as a halftone dot character. In the embodiment, since a thin character is shown as an example, the entire character is extracted as a half-dot character. Although not shown, when the character size increases, not the entire character but the edge (contour) of the character is extracted.

  A case where a light color character is present in a halftone dot having the same number of lines as in the example of FIG. 24A will be described with reference to FIG. A bright color (for example, white) character is printed in the halftone image 3020w, and when this image is subjected to processing by the adaptive smoothing processing unit 2401, a smoothed image 3030w is obtained. The halftone dot portion is smoothed and the periodic structure of the halftone dot disappears. The character portion is smoothed by the adaptive processing, and the character region 3031w remains clearly. When edge enhancement is performed on this image by the edge enhancement unit 2402, an image 3040w shown in FIG. 25B is obtained. In the image 3040w, the edge of the character (specifically, the edge on the halftone dot side) is emphasized. When an edge-enhanced image is subjected to threshold determination processing by the threshold determination unit 2403, a halftone dot character signal 3050w is obtained. A black part is an area determined as a halftone dot character. When a character brighter than the smoothed halftone dot region is arranged in the halftone dot, an edge portion on the halftone dot side is extracted as an edge portion representing the outline of the character.

A-3-4. Description of Attribute Flag Generation Unit 1006 The attribute flag generation unit 1006 (FIG. 2) includes a character determination signal input from the character determination unit 1003, a halftone dot determination signal input from the halftone dot determination unit 1004, and a character within a halftone dot. Based on the halftone dot character determination signal input from the determination unit 1005, an attribute flag is generated for each pixel. The attribute flag to be generated is determined as follows.
(1) When the halftone dot determination signal is “1”, the character determination signal is “0”, and the halftone dot character determination signal is “0”, the attribute flag is “halftone”. The halftone dot flag 1008 is turned on, and the character flag 1007 and the halftone dot character flag 1009 are turned off.
(2) When the halftone dot determination signal is “0” and the character determination signal is “1”, the attribute flag is determined to be a value indicating “character”, the character flag 1007 is turned ON, The halftone dot flag 1008 and the halftone dot character flag 1009 are OFF.
(3) When the halftone dot determination signal is “1” and the halftone dot character determination signal is “1”, the attribute flag is determined as a value indicating “halftone dot character”. The dot character flag 1009 is turned on, and the character flag 1007 and the halftone flag 1008 are turned off.
(4) In cases other than the above, the attribute flag is determined to be a value indicating “natural image, photographic image, gradation image”, and the character flag 1007, the halftone dot flag 1008, and the halftone dot character flag 1009 are all OFF. It becomes.

  In this way, flag data indicating an attribute flag for each pixel is generated. By using the flag data, various images such as an edge enhancement amount, a color reproduction method, an image forming method, and the like can be used according to the attribute for each pixel. The processing method can be controlled.

A-3-5. Description of Print Processing Based on the above, specific print processing will be described. As described above, the processed data generated from the scan data by the second input processing unit 104 (FIG. 1) and the flag data generated by the region separation processing unit 103 are compressed and stored in the main memory 113. The What will be described here is print processing executed using compressed processed image data and compressed flag data.

  The data decoding unit 109 reads the compressed processed image data and the compressed flag data stored in the main memory 113 in accordance with the timing when the printing execution unit 140 becomes ready for printing, and performs a decoding process. The decoded processed image data is stored in the second image memory 116, and the decoded flag data is stored in the second flag memory 115.

  The output processing unit 110 generates print data based on the processed image data stored in the second image memory 116 and the flag data stored in the second flag memory 115 and outputs the print data to the print execution unit 140. To do.

  FIG. 26 is a block diagram illustrating a configuration of the output processing unit 110. When the second image memory 116 and the second flag memory 115 store processed image data and flag data of a preset data amount that can be printed, the processed image data and flag data are output to the output processing unit. 110.

  The processed image data (RGB image data) transferred from the second image memory 114 is converted into CMYK image data by the RGB → CMYK conversion units 601 and 602. The CMYK image data includes a color system including color components corresponding to a color material used for printing, and in this embodiment, C (cyan), M (magenta), Y (yellow), and K (black) components. This is image data indicating the color of each pixel with the CMYK color system color values (also referred to as CMYK values). The difference between the RGB → CMYK conversion units 601 and 602 is that the former performs conversion for character images, and the latter performs conversion for photographic images and halftone dots.

  Normally, the color of characters printed in a document is black, white, or several colors at most. Therefore, in this embodiment, the RGB → YMCK conversion unit 601 converts the color to the closest color among the predefined color (C, M, Y, K) patterns. For example, when R≈G≈B≈0, it may be determined that the pixel of interest is black, so the pattern is converted into a pattern of C = M = Y = 0 and K = 255. If R≈G≈B≈255, it may be determined that the pixel of interest is white, so the pattern is converted to a pattern of C = M = Y = K = 0. The RGB → YMCK conversion unit 601 inputs the upper few bits (may be about 2 bits) when the input R, G, and B are expressed in 8 bits, and CMYK (each 8 bits) This is performed by using a look-up table (LUT) that outputs the data.

  The RGB → YMCK conversion unit 602 performs conversion from RGB → YMCK with high accuracy. This conversion is performed by matrix calculation, for example. Alternatively, it is performed using a LUT with a total of 24-bit address input for RGB and a 32-bit output for YMCK.

  The combining unit 603 combines the CMYK image data output from the two conversion units 601 and 602 based on the flag data transferred from the second flag memory 115. Specifically, when the attribute flag of the target pixel included in the flag data indicates “character”, the CMYK value of the target pixel output from the RGB → CMYK conversion unit 601 is selected and output. When the attribute flag of the target pixel indicates “natural image” or “halftone dot”, the CMYK value of the target pixel output from the RGB → CMYK conversion unit 602 is selected and output. When the attribute flag of the target pixel indicates “character in halftone dot”, a value obtained by combining the two CMYK values output from the two conversion units 601 and 602 according to a predetermined weight ( For example, an average value) is generated and output.

  A signal indicating the CMYK value of each pixel output from the combining unit 603 is supplied to the filter processing units 604 to 606. These filter processing units 604 to 606 have buffers for several lines inside and perform two-dimensional filter processing. The difference between the filter processing units 604 to 606 is that the coefficients for determining the degree of edge enhancement in the filter process are different.

  The filter processing unit 604 performs filter processing suitable for “character”. The filter processing unit 605 performs filter processing suitable for “characters in halftone dots”. The filter processing unit 606 performs filter processing suitable for “photographs” or “halftone dots”. The degree of edge enhancement of each filter processing unit is in the order of filter processing unit 604> filter processing unit 605> filter processing unit 606. However, the degree of edge enhancement is merely an example and is not necessarily limited to this.

  In other words, the strength of the smoothing process is in the order of filter processing unit 604 <filter processing unit 605 <filter processing unit 606.

  The selector 607 selects and outputs the processed CMYK value output from the filter processing unit 604 when the attribute flag of the target pixel indicates “character”. The selector 607 selects and outputs the processed CMYK value output from the filter processing unit 605 when the attribute flag of the target pixel indicates “character in halftone dot”. The selector 607 selects and outputs the processed CMYK value output from the filter processing unit 606 when the attribute flag of the target pixel indicates “photograph” or “halftone dot”.

  The processed CMYK values output from the selector 607 are input to the gamma correction units 608 and 610. The gamma correction unit 608 performs correction suitable for “character” and “character in halftone dot”, and the gamma correction unit 610 performs correction suitable for “halftone dot” and “photograph”.

  The corrected CMYK value output from the gamma correction unit 608 is input to the error diffusion processing unit 609. The error diffusion processing unit 609 converts the input CMYK value into a dot value indicating a dot formation state by halftone processing according to the error diffusion method, and outputs the dot value. In the case of characters, it is desirable that the dots generated at the time of printing are difficult to disperse. Therefore, it is preferable to use an error diffusion method for the halftone process.

  The corrected CMYK value output from the gamma correction unit 610 is input to the dither processing unit 611. The dither processing unit 611 converts the input CMYK value into a dot value indicating a dot formation state by halftone processing according to a dither method using a dither matrix, and outputs the dot value. In the case of a photographic image or halftone dot, gradation is emphasized, and therefore dithering is preferably used for halftone processing.

  The selector 612 selects the dot value output from the error diffusion processing unit 609 and outputs it to the print execution unit 140 when the attribute flag of the target pixel indicates “character” or “character in halftone dot”. Further, when the attribute flag of the target pixel indicates “photograph” or “halftone dot”, the selector 612 selects the dot value output from the dither processing unit 611 and outputs it to the print execution unit 140.

  As described above, by appropriately extracting characters, halftone dots, and character areas in halftone dots, it is possible to perform image processing according to the attributes. Accordingly, it is possible to reproduce characters clearly and photographs smoothly. In particular, in a document including halftone dots, the halftone dots and characters can be identified, and the halftone dots can be smoothed without applying a smoothing process to the characters. Therefore, suitable moire removal is possible. As a result, it is possible to print an image with high gradation by performing smoothing processing on a photograph in a document including halftone dots, removing moire, and performing dither processing. Also, clear and easy-to-read characters can be printed on characters in a document including halftone dots by performing an edge enhancement process and then an error diffusion system process.

  Further, according to the first embodiment described above, the first input processing unit 102 (FIG. 1) acquires scan data that is input image data representing an input image to be processed. Then, the luminance data acquisition unit 1002 of the region separation processing unit 103 (FIG. 2) uses the image data from the first input processing unit 102 to obtain a plurality of luminance values that are degrees of brightness of the plurality of pixels. The luminance image data including is acquired. The adaptive smoothing processing unit 2401 of the halftone dot character determination unit 1005 (FIG. 16) uses the luminance image data from the luminance data acquisition unit 1002, and the target pixel is a first type specific region in which a plurality of pixels are continuous. And whether or not the target pixel is included in a second type of specific area in which a plurality of pixels are continuous (where the brightness of the first type of specific area is the second type Dark compared to the brightness of a specific area of the species). Specifically, as described with reference to FIGS. 18 to 20, the dark line determination unit 2503a of the adaptive smoothing processing unit 2401 (FIG. 17) has the pixel of interest as the block 2703 in FIG. 19A or FIG. A) to determine whether or not a plurality of low-luminance pixels are included in a continuous specific area as in blocks 2813, 2823, and 2833 in FIG. 20C (in the example of the first type of specific area) is there). Further, as described with reference to FIGS. 18, 21, and 22, the bright line determination unit 2503b of the adaptive smoothing processing unit 2401 (FIG. 17) has the pixel of interest as the block 2703w in FIG. As in blocks 2813w, 2823w, and 2833w in FIGS. 22A to 22C, it is determined whether or not a plurality of high-luminance pixels are included in a specific area that continues (example of a second type of specific area). Is). Further, the smoothing processing unit 2502 and the selector 2504 of the adaptive smoothing processing unit 2401 (FIG. 17) determine the luminance value of the pixel (here, the dark line pixel) determined to be included in the first type specific area. The luminance values of the pixels that are determined to be included in the two types of specific regions (here, bright line pixels) are not smoothed and are not determined to be included in the first type of specific region, and the second By smoothing luminance values of pixels that are not determined to be included in the specific region of the seed, an adaptive smoothing signal 2505 (also referred to as smoothed data) that is a smoothed signal is generated. Then, the edge enhancement unit 2402 and the threshold determination unit 2403 of the halftone character determination unit 1005 (FIG. 16) use the smoothed data from the adaptive smoothing processing unit 2401 to represent edge pixels representing edges in the input image. Is detected.

  Thus, the luminance value of the pixel (here, the dark line pixel) determined to be included in the first type specific area and the pixel (here, the bright line) determined to be included in the second type specific area. Since the edge pixel is detected without smoothing the luminance value of the pixel), the pixel representing the edge of the character represented by the dark color (that is, the low luminance color) in the light background, and the dark background Both pixels representing the edge of a character represented by a bright color (that is, a high-luminance color) can be appropriately detected.

  In this embodiment, the dark line determination unit 2503a uses the above conditions (1) to (3) and the conditions (4) to (6) as conditions for detecting dark line pixels, and a bright line determination unit. 2503b uses the above conditions (1), (2), and (7) and the conditions (4), (5), and (8) as conditions for detecting bright line pixels. When these conditions are used, the first type of specific area to which the dark line pixels belong, such as a block 2703 in FIG. 19A and a block 2703w in FIG. In addition, it is possible to use a darker area than the second type specific area to which the bright line pixel belongs, and the second type specific area is an area where a plurality of pixels are continuous, An area that is brighter than one type of specific area can be used. Note that a region in which a plurality of dark line pixels detected by the dark line determination unit 2503a is continuous corresponds to at least a part of the first type specific region including the dark line pixels, and a plurality of bright lines detected by the bright line determination unit 2503b. The region where the line pixels are continuous corresponds to at least a part of the second type specific region including the bright line pixels. When the average luminance of the region where the plurality of dark line pixels are continuous is lower than the average luminance of the region where the plurality of bright line pixels are continuous, the first type of specific region is the second type of specific region and It can be said that the area is darker than that of the first type of specific area, and the second type of specific area is brighter than the first type of specific area.

  Usually, characters are often displayed in a dark color on a light background, and characters are often displayed in a light color on a dark background. However, when assessing the quality of image processing, the user is also assured that the image quality when characters are represented in light colors on a dark background is the same as the image quality when characters are represented in dark colors on a light background. It may be important. Therefore, as in this embodiment, both the edge of the character represented by a dark color on a light background and the edge of the character represented by a light color on a dark background are detected and used for image processing. As a result, high-quality image processing can be realized.

B. Second embodiment:
FIG. 27 is a flowchart illustrating another example of the process of detecting the pixels of the first type of specific area and the pixels of the second type of specific area. This process is performed by the dark line determination unit 2503a and the bright line determination unit 2503b of the adaptive smoothing processing unit 2401 (FIG. 17). The difference from the first embodiment described above is that a plurality of filters associated with lines having different thicknesses are used. In the first embodiment, as shown in FIG. 18, although the extending direction of the lines is different between the plurality of filters (specifically, the plurality of line patterns 2600 to 2630), the line thickness is approximately the same. is there. In this case, depending on the thickness of the line included in the input image, it may not be possible to make an appropriate determination in the examples of FIGS. For example, a pixel of interest on a thick line segment may not be determined to be a line pixel (dark line pixel or bright line pixel) by any of the filters shown in FIGS. Therefore, in the second embodiment, the dark line determination unit 2503a and the bright line determination unit 2503b use a plurality of filters associated with lines having different thicknesses.

  FIG. 28 is an explanatory diagram illustrating an example of a plurality of filters. 28A shows a filter 2660 having a line width Wd of 1 pixel, FIG. 28B shows a filter 2670 having a line width Wd of 2 pixels, and FIG. 28C shows a line width Wd. Shows a filter 2680 having three pixels, and FIG. 28D shows a filter 2690 having a line width Wd of four pixels. Each of these filters 2660 to 2690 shows a line pattern for determining a vertical line, similarly to the line pattern 2600 of FIG.

The filter 2660 in FIG. 28A includes three blocks 2662, 2663, and 2664 of 19 vertical pixels × 1 horizontal pixel. These blocks 2662, 2663 and 2664 are arranged side by side with a gap having the same width as the line width Wd. The pixel of interest 2661 is arranged at the center of the central block 2663.

  Similarly, the filter 2670 in FIG. 28B includes three blocks 2672, 2673, and 2675 of 19 vertical pixels × 2 horizontal pixels. The filter 2680 in FIG. 28C includes three blocks 2682, 2683, and 2684 of 19 vertical pixels × 3 horizontal pixels. The filter 2690 in FIG. 28D includes three blocks 2692, 2693, and 2694, each having 19 vertical pixels × 4 horizontal pixels. In any of the filters 2670 to 2690, the three blocks are arranged side by side with a gap having the same width as the line width Wd. The target pixels 2671, 2681, and 2691 are arranged at the center of the central blocks 2673, 2683, and 2663 (when the line width Wd is an even number, the target pixel is arranged at the pixel position on the left side of the center of the block. Have been).

  Although not shown, a plurality of usable filters include a filter indicating a line pattern for determining a horizontal line and two directions, like the filters 2610 to 2630 in FIGS. 18B to 18D. And a filter showing two types of line patterns for determining a diagonal line. A plurality of filters having different line widths can be used as each of the four types of filters having different line extending directions. In any filter, the three blocks are arranged side by side with a gap having the same width as the line width in a direction perpendicular to the extending direction of the blocks. The target pixel is arranged at the center of the central block.

  The dark line determination unit 2503a (FIG. 17) and the bright line determination unit 2503b use each filter to determine whether the target pixel is a line pixel, as in the case of using the filters 2600 to 2630 in FIG. . The dark line determination unit 2503a sets the above conditions (1) to (3) and conditions (4) to (6) to three blocks of each filter (for example, three blocks 2661 and 2663 of the filter 2660 in FIG. 28). , 2664) to evaluate and determine whether the pixel of interest is a dark line pixel located on a dark line. Similarly, the bright line determination unit 2503b evaluates the above conditions (1), (2), and (7) and the conditions (4), (5), and (8), so that the target pixel is located on a light color line. It is determined whether the pixel is a bright line pixel. The threshold value 1 and the threshold value 2 of each condition may be experimentally determined in advance so that an appropriate determination can be made.

  The dark line determination unit 2503a (FIG. 17) and the bright line determination unit 2503b determine whether the pixel of interest is a line pixel according to the procedure of FIG. In the example of FIG. 27, the dark line determination unit 2503a and the bright line determination unit 2503b cooperate to output a line switching signal. In the figure, the process surrounded by the frame denoted by reference numeral 2503a indicates the process of the dark line determination unit 2503a, and the process surrounded by the frame denoted by reference numeral 2503b indicates the process of the bright line determination unit 2503b. . S190 and S195 can be executed by either the dark line determination unit 2503a or the bright line determination unit 2503b. Note that the dark line determination unit 2503a and the bright line determination unit 2503b may independently output a dark line switching signal and a bright line switching signal, as in the first embodiment.

  In S100, the dark line determination unit 2503a initializes the line switching flag to OFF. In S105, the dark line determination unit 2503a initializes the current line width Wd to the minimum value (here, 1) of the available filter line widths Wd. In S110, the dark line determination unit 2503a selects a filter associated with the current line width Wd. As described above, in this embodiment, four filters having different line extending directions are associated with one line width Wd. In S110, four filters are selected. In S115, the dark line determination unit 2503a uses the four filters selected in S110 to determine whether the target pixel is a dark line pixel for each filter. In S120, the dark line determination unit 2503a determines whether or not the target pixel is determined to be a dark line pixel by at least one of the four filters.

  When it is determined that the target pixel is a dark line pixel by at least one filter (S120: Yes), in S190, the dark line determination unit 2503a sets the line switching flag to ON. In S195, the dark line determination unit 2503a outputs a line switching signal (here, “1” line switching signal) associated with the line switching flag to the selector 2504. Then, the process of FIG. 27 ends.

  If any of the four filters is used and the target pixel is not determined to be a dark line pixel (S120: No), the dark line determination unit 2503a performs processing of all available line widths Wd in S125. It is determined whether or not it has been completed. In the example of FIG. 27, it is determined whether or not the current line width Wd is greater than or equal to the maximum value WdMax of the available filter line width Wd. In the present embodiment, the maximum value WdMax is 4 as described in FIG.

  When the unprocessed line width Wd remains (S125: No), in S130, the dark line determination unit 2503a updates the current line width Wd to a line width Wd that is one step larger (in this embodiment, the line width 1 is added to Wd). Then, the dark line determination unit 2503a proceeds to S110 and executes the process of the updated line width Wd.

  When the processing for all the line widths Wd is completed (S125: Yes), the processing proceeds to S155.

  In S155, the bright line determination unit 2503b initializes the current line width Wd to the initial value Wdiw for the bright line. The bright line initial value Wdiw is set to a value larger than the dark line initial value (here, 1). The bright line determination unit 2503b makes a determination using a filter having a line width Wd equal to or greater than the initial value Wdiw, without using a filter having a line width Wd less than the initial value Wdiw for the bright line. The reason for this is as follows. When the width of a bright color line is thin, the thin line is likely to be crushed in a printed image due to bleeding or misalignment of a color material representing a dark ground around the line. When the thin line is crushed, the visibility of the line is lowered, and the effect of improving the image quality of the image processing for the characters in the halftone dot is reduced. On the other hand, when the width of the dark color line is narrow, the line can be easily visually recognized even in the case where bleeding or displacement of the color material representing the dark color line occurs in the printed image. Therefore, the effect of improving the image quality of the image processing for the characters in the halftone dot is high. As described above, the minimum line width Wd (here, 1) of the filter used in the dark line processing is smaller than the minimum line width Wdiw of the filter used in the bright line processing.

  In S160, the bright line determination unit 2503b selects a filter associated with the current line width Wd. S160 is performed in the same manner as S110. In S165, the bright line determination unit 2503b uses the four filters selected in S160 to determine whether or not the target pixel is a bright line pixel for each filter. In S170, the bright line determination unit 2503b determines whether or not the target pixel is determined to be a bright line pixel by at least one of the four filters.

  When it is determined that the target pixel is a bright line pixel by at least one filter (S170: Yes), in S190, the bright line determination unit 2503b sets the line switching flag to ON. In S195, the bright line determination unit 2503b outputs a line switching signal (here, a line switching signal of “1”) associated with the line switching flag to the selector 2504. Then, the process of FIG. 27 ends.

  If it is not determined that the pixel of interest is a bright line pixel using any of the four filters (S170: No), the bright line determination unit 2503b performs processing of all available line widths Wd in S175. It is determined whether or not it has been completed. Specifically, it is determined whether or not the current line width Wd is equal to or greater than the maximum value WdMax of the available filter line width Wd.

  When the unprocessed line width Wd remains (S175: No), in S180, the bright line determination unit 2503b updates the current line width Wd to a line width Wd that is one step larger (in this embodiment, the line width Wd 1 is added to the width Wd). Then, the bright line determination unit 2503b proceeds to S160 and executes the process of the updated line width Wd.

  When all the line widths Wd have been processed (S175: Yes), in S195, the bright line determination unit 2503b sends a line switching signal associated with the line switching flag (here, a line switching signal of “0”). Is output to the selector 2504. Then, the bright line determination unit 2503b ends the process of FIG.

  In this embodiment, the selector 2504 (FIG. 17) receives the line switching signal output in S195 from either the dark line determination unit 2503a or the bright line determination unit 2503b. Then, when the line switching signal is “1”, that is, when the dark line determination unit 2503a determines that the pixel of interest is a dark line pixel, the selector 2504 determines that the pixel of interest is bright in the bright line determination unit 2503b. A determination signal 2501 that is not filtered is output at each time when it is determined that the pixel is a line pixel. The selector 2504 outputs a smoothed signal when the line switching signal is “0”, that is, when it is determined that the target pixel is neither a dark line pixel nor a bright line pixel.

FIG. 29 is an explanatory diagram of processing for determining the initial value Wdiw for bright lines (that is, usable filters). FIG. 29A shows a flowchart of the determination process. In S200, test pattern data corresponding to each of a plurality of filters having different line widths Wd is generated. FIG. 29B to FIG. 29E are explanatory diagrams illustrating examples of test patterns represented by test pattern data. FIGS. 29B to 29E show test patterns TP1, TP2, TP3, and TP4 for Wd = 1, 2, 3, and 4, respectively. Each test pattern TP1 to TP4 represents a black base BG and white vertical lines WL1 to WL4, respectively. The white vertical lines WL1 to WL4 are represented by data representing white vertical lines having a corresponding line width Wd. For example, the white vertical line WL1 of the test pattern TP1 for Wd = 1 is represented by data representing a white vertical line having a line width of 1. Any method can be adopted as a method for generating test pattern data. In this embodiment, a computer (not shown) generates test pattern data according to a program.

  In step S210, the control unit 120 of the multifunction peripheral 100 causes the print execution unit 140 to print the test patterns TP1 to TP4 using the test pattern data. In S220, the user determines whether the white line can be identified by visually observing the printed test patterns TP1 to TP4. As described above, when the line width of the printed white line is small, the white line may be crushed due to bleeding of the color material. When the thin white line is crushed, the user cannot identify the white line. Then, the user specifies the thinnest white line among the identifiable white lines. The line width Wd associated with the identified white line is referred to as the minimum width Wmin.

  In S230, the initial value Wdiw for the bright line is set to the minimum width Wmin. Data representing the set minimum width Wmin is incorporated in the bright line determination unit 2503b (FIG. 17) (for example, data representing the minimum width Wmin is written in a non-illustrated nonvolatile memory of the bright line determination unit 2503b). . Then, the process of FIG. 29A ends. Accordingly, the bright line determination unit 2503b can appropriately detect pixels of a bright color line that can be identified when printed.

  As described above, in the second embodiment, the dark line determination unit 2503a and the bright line determination unit 2503b are filters for detecting pixels representing lines, and are associated with lines having different thicknesses. Is used to determine whether or not the target pixel is included in a specific region representing a line. Accordingly, it is possible to appropriately detect pixels that represent the edges of characters represented by lines of various thicknesses.

  Further, the initial value Wdiw of the line width Wd for bright lines is larger than the initial value (here, 1) of the line width Wd for dark lines. Therefore, the total number P of filters that can be used by the dark line determination unit 2503a is larger than the total number Q of filters that can be used by the bright line determination unit 2503b (in this embodiment, the total number P = 16 (total number 4 of the line width Wd 4 × Number of types of X-ray directions 4), the total number Q is 1 or more and an integer different from P). Thereby, it is possible to appropriately detect the edge of a thin character represented by a continuous line of low luminance pixels. In addition, it is possible to appropriately detect the edge of a character represented by a continuous line of high luminance pixels.

  The line width Wd (here, 1) of the thinnest line among the plurality of lines associated with the plurality of filters used by the dark line determination unit 2503a corresponds to the plurality of filters used by the bright line determination unit 2503b. It is narrower than the line width Wdiw of the thinnest line among the plurality of lines to be attached. Therefore, it is possible to appropriately detect an edge of a character represented by a thin line with continuous low-luminance pixels.

  In addition, the region separation processing unit 103 (FIG. 1), the second input processing unit 104, and the output processing unit 110 add the input image data acquired by the first input processing unit 102 within the halftone dots of the region separation processing unit 103. By performing image processing using the edge pixels detected by the character determination unit 1005 (FIG. 16), print data that is image data for printing is generated. Then, the output processing unit 110 supplies the generated print data to the print execution unit 140. Here, as described with reference to FIG. 29A, the line width Wdiw of the thinnest line among the plurality of types of lines associated with the plurality of bright line filters is the color used by the print execution unit 140. It is determined in advance based on the material blur identification. Therefore, it is possible to detect the edge of a character represented by a line having a thickness suitable for the bleeding characteristic of the color material used by the print execution unit 140.

C. Variation:
(1) The conditions for determining that the target pixel is included in the first type specific area (relatively dark area) by the dark line determination unit 2503a (FIG. 17) are the above-described conditions (1) to (3). Not only the conditions (4) to (6) but various conditions may be used. Similarly, the conditions for determining that the target pixel is included in the second type specific area (relatively bright area) by the bright line determination unit 2503b are the conditions (1), (2), and (7) described above. (4) Not limited to (5) and (8), various conditions may be used. For example, the condition for determining that the target pixel is included in the first type of specific region is that the average value of the luminance values of the pixel block including the target pixel (for example, the block 2603 in FIG. 18A) is It may include being smaller than a predetermined first threshold value (for example, 128). Similarly, the condition for determining that the target pixel is included in the second type specific region is that the average value of the luminance values of the pixel block including the target pixel (for example, the block 2603 in FIG. 18A) is The predetermined threshold value may be greater than a predetermined second threshold value (for example, 128) (the second threshold value is equal to or greater than the first threshold value).

  In general, the condition for determining that the target pixel is included in the first type of specific area (referred to as a dark area condition) is a plurality of pixels whose average luminance is lower than that of the second type of specific area. There may be various conditions indicating that the pixel of interest is included in the first type specific area that is a continuous area. The condition for determining that the pixel of interest is included in the second type specific area (referred to as a bright area condition) is that a plurality of pixels having higher average luminance than the first type specific area are continuous. There may be various conditions indicating that the target pixel is included in the second type specific area which is an area. For example, as the dark area condition and the bright area condition, various conditions configured such that the average luminance value in the first type specific area is darker than the average luminance value in the second type specific area are adopted. You can do it.

(2) One or more filters that can be used by the dark line determination unit 2503a and one or more filters that can be used by the bright line determination unit 2503b are other than the filters described in FIGS. Various filters may be used. For example, the interval between three blocks included in one filter (for example, the three blocks 2662, 2663, and 2664 in FIG. 28A) may be different from the width of each block. In general, as a filter, it is possible to detect various filters that can detect pixels that represent a line of a specific thickness, that is, pixels that are included in an area that represents a line of a specific thickness. Various filters can be employed.

  In addition, the total number Q of filters that can be used by the bright line determination unit 2503b may be the same as the total number P of filters that can be used by the dark line determination unit 2503a. Instead of this, an arbitrary integer different from the total number P is used. It may be. Here, P> Q may be sufficient and P <Q may be sufficient.

  The minimum line width of one or more filters for dark lines that can be used by the dark line determination unit 2503a is larger than the minimum line width of one or more filters for bright lines that can be used by the bright line determination unit 2503b. Also good. Further, the minimum line width of one or more filters that can be used by the bright line determination unit 2503b may be determined independently of the bleeding characteristics of the color material.

  Further, as a filter, it is possible to determine whether or not there is a difference in luminance value between a linear region extending in a specific direction and at least one of both regions sandwiching the linear region. Various filters may be employed.

  Note that the dark line determination unit 2503a may determine whether or not the target pixel is included in the first type specific region without using such a filter. For example, the dark line determination unit 2503a identifies an area where pixels having a luminance value less than a predetermined dark threshold by analyzing the luminance image data, and adopts the identified area as the first type specific area. Also good. Similarly, the bright line determination unit 2503b may determine whether or not the target pixel is included in the second type specific region without using such a filter. For example, the bright line determination unit 2503b identifies an area where pixels whose luminance value exceeds a predetermined bright threshold value continues by analyzing the luminance image data, and adopts the identified area as the second type specific area. (Here, the bright threshold is equal to or higher than the dark threshold).

(3) In each of the above embodiments, the processed image data is used by the print execution unit 140 to print the processed image (FIG. 26). However, the processed image data may be used to display the processed image on a display unit such as a liquid crystal display.

(4) In each of the above embodiments, the input signal 1001 (FIG. 1) is an RGB value. Instead, color values of other color systems may be used. For example, the value of each pixel of the scan data may be input to the region separation processing unit 103 as the input signal 1001 after being converted into a CMY color system color value.

(5) Specific processing executed by each unit included in the region separation processing unit 103 of each of the above embodiments, for example, the character determination unit 1003, the halftone dot determination unit 1004, and the halftone dot character determination unit 1005 is an example. , As appropriate. For example, the peripheral range of vertical 3 pixels × horizontal 3 pixels used by the area integration units 1105 and 1106 in FIG. 3 is a range of other sizes, for example, a range of vertical 3 pixels × horizontal 5 pixels or vertical 5 pixels × horizontal 5 pixels. It may be. Further, the integration processing unit 1209 in FIG. 12 uses three types of integration ranges, but the size and the number of types of these integration ranges can be changed as appropriate. Further, the threshold for the threshold determination unit 1211 to determine the integration result is appropriately adjusted according to the size of the integration range to be used. When the overall determination unit 1213 in FIG. 12 generates a halftone determination signal using a plurality of determination results output for each integration range by the threshold determination unit 1211, for example, logical sum, logical product The combination of these logical operations can be changed as appropriate.

(6) The process (FIG. 26) in which the output processing unit 110 generates print data is an example and can be changed as appropriate. For example, although the output processing unit 110 has shown an example in which the error diffusion method and the dither method are employed as the halftone processing, the present invention is not limited thereto. For example, when only the error diffusion method is used, the error diffusion matrix size of the former is reduced and the error diffusion matrix size of the latter is increased for characters (including characters in halftone dots) and photographs (including halftone dots). Is preferred. When only dither processing is used, different dither matrix patterns may be used for characters and photographs. Also, other methods may be employed as the halftone processing method.

(7) In the above embodiment, the input image data is scan data, but is not limited thereto. The input image data may be captured image data generated by optically photographing a document such as a printed matter with a digital camera including a two-dimensional image sensor. Further, it may be image data generated using an application program such as a word processor.

(8) The edge pixels detected by the edge enhancement unit 2402 and the threshold determination unit 2403 may be used for arbitrary processing. For example, it may be used to generate edge image data representing an edge.

(9) The processing executed by the multifunction device 100 of FIG. 1, for example, all or part of the processing executed by the area separation processing unit 103, the second input processing unit 104, and the output processing unit 110 is not limited to the multifunction device 100. May be executed by various devices. For example, a scanner or a digital camera may execute processing executed by the region separation processing unit 103 or the second input processing unit 104 using input image data generated by itself. Further, for example, a terminal device (not shown) or a server (not shown) connected to be communicable with a scanner or a printer uses the scan data acquired from the scanner, and the region separation processing unit 103 and the second input processing unit 104. The processing executed by the output processing unit 110 may be executed. Then, print data generated as a result may be supplied to the printer. In addition, a plurality of computers (for example, cloud servers) that can communicate with each other via a network share part of the processing executed by the region separation processing unit 103, the second input processing unit 104, and the output processing unit 110. As a whole, image processing may be executed. In this case, the entirety of the plurality of computers is an example of the image processing apparatus.

  In each of the above embodiments, all or part of the processing executed by the region separation processing unit 103, the second input processing unit 104, and the output processing unit 110 is executed by these hardware circuits, but is a CPU. It may be executed by the main processor 101. In this case, a program for processing executed by the main processor 101 among processing executed by the region separation processing unit 103, the second input processing unit 104, and the output processing unit 110 is included in the computer program PG. For example, in the first embodiment, the main processor 101 uses the process of generating luminance image data executed by the area separation processing unit 103 and the luminance image data, so that the target pixel is included in the first type specific area. Processing to determine whether or not the target pixel is included in the second type specific area using the luminance image data, and smoothing the luminance image data using the determination result By executing the computer program PG, the processing for generating the smoothed data and the processing for detecting the edge pixels representing the edges in the input image using the smoothed data may be realized. .

  As described above, in each of the above embodiments, a part of the configuration realized by hardware may be replaced with software. Conversely, part or all of the configuration realized by software is replaced with hardware. You may make it replace.

  As mentioned above, although this invention was demonstrated based on the Example and the modification, Embodiment mentioned above is for making an understanding of this invention easy, and does not limit this invention. The present invention can be changed and improved without departing from the spirit and scope of the claims, and equivalents thereof are included in the present invention.

DESCRIPTION OF SYMBOLS 100 ... MFP, 101 ... Main processor, 102 ... First input processing unit, 103 ... Region separation processing unit, 104 ... Second input processing unit, 106 ... Interpreter, 107 ... Data compression unit, 108 ... Resolution conversion unit, 109 DESCRIPTION OF SYMBOLS ... Data decoding part, 110 ... Output processing part, 111 ... 1st image memory, 112 ... 1st flag memory, 113 ... Main memory, 114 ... 2nd image memory, 115 ... 2nd flag memory, 116 ... 2nd image memory DESCRIPTION OF SYMBOLS 117 ... Nonvolatile memory, 120 ... Control part, 130 ... Reading execution part, 140 ... Print execution part, 200 ... Multifunction machine, 601 ... Conversion part, 603 ... Composition part, 604 ... Filter processing part, 604 ... Filter processing part 605: Filter processing unit, 606: Filter processing unit, 607 ... Selector, 608 ... Gamma correction unit, 609 ... Error diffusion processing unit, 610 ... Gamma correction unit, 611... Dither processing unit, 612... Selector, 1001... Input signal, 1002 .. Luminance data acquisition unit, 1003... Character determination unit, 1004. Attribute flag generation unit, 1007 ... Character flag, 1008 ... Halftone dot flag, 1009 ... Halftone dot character flag, 1102 ... Edge emphasis unit, 1103 ... Threshold decision unit, 1104 ... Threshold decision unit, 1105 ... Area integration unit, 1106 ... Area integration unit, 1107 ... threshold judgment unit, 1108 ... threshold judgment unit, 1109 ... total judgment unit, 1110 ... character judgment signal, 1202 ... edge enhancement unit, 1203 ... threshold judgment unit, 1204 ... threshold judgment unit, 1205 ... isolation amount Determination unit, 1206 ... isolated amount determination unit, 1209 ... integration processing unit, 1211 ... threshold determination unit, 1213 ... comprehensive determination unit, 12 5 ... halftone dot determination signal, 1302 ... outer edge area, 1303 ... inner edge area, 1304 ... area, 1305 ... area, 1306 ... area, 1501 ... edge boundary, 1502 ... character, 1503 ... background, 1504 ... luminance image data , 1507 ... edge extraction data, 1510 ... cross section, 1601 ... halftone edge boundary part, 1601 ... edge boundary part, 1602 ... halftone dot, 1603 ... background, 1604 ... luminance image data, 1607 ... edge extraction data, 1610 ... cross section, 1700 ... inner edge signal, 1710 ... determination pattern, 1712 ... black pixel, 1713 ... white pixel, 1810 ... inner edge signal, 1820 ... inner edge signal, 1830 ... inner edge signal, 1840 ... inner edge signal, 2011 ... first integration , 2012... Second accumulator, 2013... Third accumulator, 2021. 22 ... 2nd determination unit, 2023 ... 3rd determination unit, 2101 ... Pixel of interest, 2102 ... Integration range, 2103 ... Integration range, 2104 ... Integration range, 2205 ... Edge, 2401 ... Adaptive smoothing processing unit, 2402 ... Edge enhancement , 2403 ... threshold value determination unit, 2501 ... determination signal, 2502 ... smoothing processing unit, 2503a ... dark line determination unit, 2503b ... bright line determination unit, 2504 ... selector, 2505 ... adaptive smoothing signal, PG ... computer program

Claims (7)

An image processing apparatus,
An input image data acquisition unit for acquiring input image data representing the input image;
Using the input image data, a luminance image data acquisition unit that acquires luminance image data including a plurality of luminance values that are degrees of brightness of a plurality of pixels;
A first determination unit that determines whether or not a target pixel is included in a first type of specific region using the luminance image data, wherein the first type of specific region is a region in which a plurality of pixels are continuous. The first type specific area is an area that is darker than the second type specific area; and
A second determination unit configured to determine whether the target pixel is included in the second type of specific region using the luminance image data, wherein the second type of specific region includes a plurality of pixels; The second determination unit, wherein the second type specific region is a brighter region than the first type specific region;
A smoothed data generation unit that generates smoothed data by smoothing the brightness image data, and the brightness value of a pixel that is determined to be included in the first type specific region among the plurality of pixels And the luminance value of the pixel determined to be included in the second type of specific area are not determined to be included in the first type of specific area without being smoothed, and the second type of specific value is determined. The smoothed data generation unit that generates the smoothed data by smoothing luminance values of pixels that are not determined to be included in the region;
A detection unit that detects edge pixels representing edges in the input image using the smoothed data;
An image processing apparatus comprising: The image processing apparatus according to claim 1,
At least one of the first determination unit and the second determination unit is a filter for detecting a pixel representing a line, and using a plurality of filters associated with lines having different thicknesses, Determining whether the pixel of interest is included in the specific region representing a line;
Image processing device. The image processing apparatus according to claim 2,
The first determination unit is a filter for detecting pixels representing a line, and uses at least one of P (P is an integer of 1 or more) filters associated with lines having different thicknesses. Determining whether the pixel of interest is included in the first type specific region representing a line;
The second determination unit is a filter for detecting pixels representing a line, and is Q (Q is an integer different from P) that is associated with lines having different thicknesses. Using at least one of the filters to determine whether the pixel of interest is included in the specific region of the second type representing a line;
Image processing device. The image processing apparatus according to claim 3,
P> Q,
Image processing device. The image processing apparatus according to claim 4,
The width of the narrowest line among the P types of lines associated with the P filters that can be used by the first determination unit is the Q associated with the Q filters that can be used by the second determination unit. Thinner than the width of the thinnest line of types,
Image processing device. The image processing apparatus according to claim 5,
An output data generation unit that generates output data that is image data for printing by performing image processing using the edge pixels on the input image data;
A supply unit that supplies the output data to a print execution unit that prints an image using a color material;
With
The narrowest line width among the Q types of lines associated with the Q filters that can be used by the second determination unit is determined in advance based on the bleeding characteristics of the color material used by the print execution unit. Being
Image processing device. A computer program for image processing,
An input image data acquisition function for acquiring input image data representing an input image;
Using the input image data, a luminance image data acquisition function for acquiring luminance image data including a plurality of luminance values that are degrees of brightness of a plurality of pixels,
A first determination function for determining whether or not a target pixel is included in a first type of specific region using the luminance image data, wherein the first type of specific region is a region in which a plurality of pixels are continuous. The first determination function, wherein the first type specific area is an area that is darker than the second type specific area;
A second determination function for determining whether or not the target pixel is included in the second type of specific region using the luminance image data, wherein a plurality of pixels are continuous in the second type of specific region. The second determination function, wherein the second type specific area is an area brighter than the first type specific area;
A smoothed data generation function for generating smoothed data by smoothing the brightness image data, and the brightness value of a pixel determined to be included in the first type specific region among the plurality of pixels And the luminance value of the pixel determined to be included in the second type of specific area are not determined to be included in the first type of specific area without being smoothed, and the second type of specific value is determined. The smoothed data generation function for generating the smoothed data by smoothing luminance values of pixels that are not determined to be included in the region; and
A detection function for detecting edge pixels representing edges in the input image using the smoothed data;
A computer program that causes a computer to realize

Images