JPH07212591A - Image binarizer - Google Patents

Abstract

(57) [Summary] [Purpose] Create a histogram of the brightness value of the image,
The black representative pixel value is obtained, and the binarization threshold value is obtained at high speed from the average thereof. A video signal of an original image is processed by a signal processing unit 1 and A / D converted by an A / D conversion unit 2 to obtain a pixel value a having multi-level luminance. The A / D converted pixel value a is
It is output to the histogram creation unit 3, and the histogram creation unit 3 creates a histogram 6 of the brightness values of the pixels. The representative pixel value detection unit 4 detects a black pixel representative value c and a white pixel representative value d from the histogram 6. The black pixel representative value c and the white pixel representative value d are passed to the threshold value determining unit 5, and the binarization threshold value is determined. The method of determination is to use the average of the black pixel representative value and the white pixel representative value as the binarization threshold.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image binarization device, and more particularly to an image binarization device for removing a shadow or unevenness on a multivalued image to obtain a threshold value for binarization.

[0002]

2. Description of the Related Art Conventionally, as a method of obtaining a threshold value for binarization, a temporary threshold value is determined, an evaluation function value of the binarization quality is obtained from a histogram, and the temporary threshold value is changed to obtain an evaluation function value. There are various methods, such as a method that uses a threshold when the maximum and minimum values are taken, a method that uses a fixed value, a method that obtains a threshold from the maximum and minimum pixel values, and a method that takes the median value of a histogram. It was

Conventionally, a machine for reading a document, such as a copying machine, a scanner, or a fax, has a light source inside the device,
The light emitted from the light source is applied to the document, and the reflected light is reflected by the CCD (Char
ge Coupled Device: Charge coupled device). Since the light source is constant, shadows and uneven brightness are unlikely to occur, and even if they occur, the environment inside the device is almost constant, so the way shadows and uneven brightness occur is constant, and correction is easy. . On the other hand, images captured by video cameras, electronic still cameras, etc. have various numbers of light sources, positions, and intensities, and shadows and brightness unevenness are likely to occur, and the way in which they occur is not constant. It was difficult to correct.

A known document describing a conventional image binarization device is, for example, Japanese Patent Laid-Open No. 63-95779. In this publication, in order to automatically detect the white level of the background portion and the black level of the character / graphic portion on the document, the determination means for determining the area of a predetermined image content in the image, and the determination means Means for obtaining a level expected to be black on the original image for each predetermined number of scanning lines in the image area, and the level expected to be black obtained for each of the predetermined number of scanning lines is the density of the black portion on the document. Based on the verification result by the verification means and the verification result by the verification means, the black level of the predetermined area on the image is either the value previously obtained, or the newly obtained black And a selecting means for selecting whether or not the level is replaced with an expected level.

Further, in Japanese Patent Laid-Open No. 3-237571, for binarizing multivalued images such as grayscale images and range images, a threshold value for obtaining the best binary image is calculated. Then, for each pixel in the window, the difference between the brightness and the brightness of the specific pixel is compared with a parameter proportional to the contrast of the image portion of interest, and the image in the window is binarized. In the case where a plurality of image parts having different contrasts exist in one image, by selecting the parameter according to the contrast of the target image part, an optimum binarization threshold value for the image part can be obtained. It can be calculated. Further, according to Japanese Patent Laid-Open No. 4-92982, the peak of the character / figure portion does not exist in the histogram of the brightness of the image, or the valley portion does not exist between the character / figure portion and the background portion. However, the image is binarized by automatically obtaining an appropriate binarization threshold value.

[0006]

As described above, in the conventional image binarization method, the method of using the evaluation function has to have a drawback that it takes a long processing time because the evaluation function value must be obtained every time the threshold value is changed. However, the method of using the fixed value, the maximum value, the minimum value, and the median value has drawbacks such as poor quality of the binary image. In addition, since copiers, scanners, fax machines, etc. can reliably read documents, etc., it is easy to binarize and output them, but in terms of size, weight, portability, power, etc. of the device. It is inferior to a video camera. On the other hand, video cameras and electronic still cameras are superior in terms of size, weight, portability, power consumption, etc., but even if a document or the like is photographed, shadows and uneven brightness are likely to occur, and the image is binarized and output. It had the drawback of being difficult.

The present invention has been made in view of the above situation, and a histogram of the brightness values of an image is created, and the brightness value of the histogram from the brighter side to the darker side of the histogram is represented by white and black representative pixel values. And perform high-speed and high-quality binarization by using the average as the binarization threshold, and divide the image into blocks and set the binarization threshold or the white / black pixel representative value for each block. The block without characters is interpolated from the surroundings, and the threshold of the block is set to the threshold for each pixel.
It is an object of the present invention to provide an image binarization device that removes shadows and uneven brightness in an image.

[0008]

In order to achieve the above object, the present invention provides (1) a histogram creating means for finding a histogram of luminance values from a multivalued image, and an image from the histogram found by the histogram creating means. A representative pixel value detecting means for obtaining a value representative of the white pixel value and the black pixel value, and a threshold value for binarization from the white pixel representative value and the black pixel representative value obtained by the representative pixel value detecting means. Or (2) block threshold detection means for obtaining a binarization threshold value for each block of the multi-valued image divided into blocks, and whether or not the block contains characters or figures And a threshold for binarizing a block that is determined by the character / graphic block detection unit not to include a character or graphic as a threshold for surrounding blocks. A non-character figure block threshold calculation means for calculating the threshold value, a pixel threshold calculation means for converting the binarization threshold value calculated for each block into a threshold value for each pixel, and each pixel calculated by the pixel threshold value calculation means. And binarizing means for binarizing the original multi-valued image using the binarizing threshold value, and (3) in (2), the threshold value obtained by the block threshold value detecting means. Among them, a threshold protruding block removing unit that replaces the threshold value of the block having a protruding value compared with the surrounding block with a value calculated from the threshold value of the surrounding block is further provided, and further, (4) above (2) Alternatively, in (3), with respect to the threshold for each block obtained by the non-character figure block threshold detecting means, block threshold smoothing for reducing a difference between a threshold of a certain block and a threshold of surrounding blocks And (5) in (2), (3) or (4), an edge enhancing means for applying edge enhancement to a multi-valued image to be compared with a threshold value by the binarizing means, Alternatively, it has an edge enhancing means for applying edge enhancement to the original multi-valued image in advance, or (6) obtains a representative value of white pixels and a representative value of black pixels for each block of the multi-valued image divided into blocks. Block black-and-white representative pixel value detecting means, character / graphic block detecting means for detecting whether or not a character or graphic is included in the block, and block determined by the character / graphic block detecting means to not include a character or graphic Non-character figure block representative value calculation means for calculating the representative pixel value of the non-character figure block representative value from the representative pixel values of the surrounding blocks, and a representative of white pixels and black pixels obtained by the non-character figure block representative value calculation means. Block threshold calculation means for obtaining a binarization threshold value of the block from the prime value, pixel threshold value calculation means for converting the binarization threshold value obtained for each block into a threshold value for each pixel, and the pixel threshold value calculation means Binarizing means for binarizing the original multi-valued image by using the binarization threshold value for each pixel obtained in (1), and (7) in (6), the block monochrome representative Of the representative values of the white pixel and the black pixel obtained by the pixel value detecting means, the representative pixel value of the block whose value is higher than that of the surrounding block is calculated from the representative pixel value of the surrounding block. Having a representative pixel value protruding block removing means for replacing
(8) In (6) or (7), with respect to the block determined by the character / graphic block detecting means to include no character or graphic, which of the white pixel representative value and the black pixel representative value of the block is reliable. Block black and white representative pixel value selecting means for selecting whether or not possible, and non-character graphic block representative value calculating means for calculating the representative pixel value not selected by the block black and white representative pixel value selecting means from the representative pixel values of surrounding blocks And (9) in (6), (7) or (8), white pixels for each block obtained by the non-character graphic block representative value calculating means,
A block black-and-white representative pixel value smoothing unit for reducing a difference between a representative value of a block and a representative value of surrounding blocks with respect to a representative value of black pixels;
(10) In (6), (7) or (8) above, the difference between the threshold of a certain block and the threshold of the surrounding blocks is calculated with respect to the binarization threshold for each block obtained by the block threshold calculation means. Block threshold smoothing means for reducing the size is provided, and further, (11) above (6) to (1)
In any one of 0), an edge enhancing means for applying edge enhancement to the multivalued image to be compared with the threshold value by the binarizing means, or an edge enhancing means for applying edge enhancement to the original multivalued image in advance. It is characterized by having.

[0009]

The histogram is obtained by the histogram creating means, the representative values of the white pixels and the black pixels are obtained at a higher speed than the representative pixel value detecting means using the histogram, and the binarization threshold is obtained by the threshold determining means. In a histogram such as a character image, black pixel peaks and white pixel peaks appear, but it is good to select the valley between the black pixel peaks and white pixel peaks as the binarization threshold. The result is obtained. Therefore, if the representative values of the white pixels and the black pixels are selected so as to correspond to the peaks on the histogram and the threshold value is selected from between them, a good binarization result can be obtained. Also, in the case of an image with only thin lines, only one of the black pixel peaks or the white pixel peaks produces a clear peak, and the other may be widely and thinly dispersed, but the part far from the dispersed part By taking the value of as a representative value, the threshold of binarization is less likely to be applied to the mountain portion, so that a good binarization result can be obtained. Further, since the threshold for binarization is directly obtained from the histogram without using the evaluation function, the threshold can be obtained at high speed.

Further, although the block threshold detection means can obtain the binarization thresholds of all the blocks, this threshold has no meaning for a blank sheet or a plain portion that does not include characters or figures. By the character / graphic block detection means and the non-character / graphic block threshold calculation means, appropriate threshold values can be obtained for all blocks, whether or not there are blocks with or without characters and graphics. Even if there are shadows and brightness unevenness in the entire image, if the size of the block is sufficiently smaller than the size of the shadows and brightness unevenness, there will be no shadows or brightness unevenness only by looking inside the block. , This threshold is an appropriate threshold. If there is unevenness in brightness that is smaller than the size of the block due to reflection of illumination, etc., an appropriate block threshold value or white / black representative pixel value cannot be obtained. Since the white / black representative pixel value is projected as compared with the surrounding blocks, it can be removed by the threshold protruding block removing means or the representative pixel value protruding block removing means.

Since a non-character graphic block generally has a solid color, it is often necessary to make it black or white. Therefore, when using the white / black representative pixel value to obtain the threshold value,
If one of the representative pixel values is left without being discarded, it becomes easy to follow uneven brightness. If the threshold value obtained in each block and the pixel in the block are directly compared and binarized, the boundary between blocks having different threshold values becomes conspicuous.
The pixel threshold calculation means interpolates the thresholds of the surrounding blocks for each pixel to calculate the thresholds, so that a smooth and natural binarized image can be obtained without making the block boundaries noticeable. In addition, by performing edge emphasis, thin characters and thin characters are not crushed by binarization, and binarization that is strong against threshold fluctuation can be performed. As a result, it is possible to perform binarization so that a character or the like can be recognized even in an image having a shadow or uneven brightness.

[0012]

Embodiments will be described below with reference to the drawings. FIG. 1 is a configuration diagram for explaining an embodiment (claim 1) of an image binarization device according to the present invention. In the figure, 1 is a signal processing unit, 2 is an A / D conversion unit, and 3 is a histogram. The creating unit 4, the representative pixel value detecting unit 5 and the threshold value determining unit 5.

The video signal of the original image is subjected to signal processing by the signal processing unit 1 and A / D converted by the A / D conversion unit 2 to obtain a pixel value a having multi-tone brightness. In the embodiment of the present invention,
A pixel value of 256 gradations will be described. The A / D-converted pixel value a is output to the histogram creation unit 3, and the histogram creation unit 3 creates a histogram b of the brightness values of the pixels. The representative pixel value detector 4 detects a black pixel representative value c and a white pixel representative value d from the histogram b. The black pixel representative value c and the white pixel representative value d are passed to the threshold value determining unit 5, and the binarization threshold value is determined. The simplest method of determination is to use the average of the black pixel representative value c and the white pixel representative value d as the binarization threshold.

2 and 3 are block diagrams of the representative pixel value detecting unit in FIG. 1, FIG. 2 shows a representative pixel value detecting unit for white pixels, and FIG. 3 shows a representative pixel value detecting unit for black pixels. There is. In the figure, 11 is a white register, 12 is a decoder, and 13-1 to 13
-n is a register, 14 is a comparator, 15 is an inverter, 16 is a white level counter, and 17 is a comparator.

In FIG. 2, the data b of the histogram created by the histogram creating unit 3 of FIG. 1 is stored in the registers 0 (13-1) to 255 (13-n). The initial value of the white register 11 is the maximum brightness value (255 in this case), and is given before the loop processing described below is started. The value after the loop processing is the representative value e of the white pixels.

The beginning of the loop processing starts with the decoding of the value of the white register 11. The value of the white register 11 is decoded by the decoder 12, and registers 13-1 to 13 in which the histogram value of the brightness value of the white register 11 is stored.
The control signal f is determined so that any one of 3-n is selected. The histogram value g read out from the register selected only by the control signal f is stored in the comparator 14
Sent to and compared with 0. If the histogram value g is equal to 0, the decrement signal h becomes valid by the comparator 14 and the value of the white register 11 is decremented by 1. If it is not 0, the decrement signal i is validated by the comparator 14 and the inverter 15 and sent to the histogram registers 13a to 13n and the white level counter 16.

The decrement signal i sent to the histogram register is valid only for the register selected by the control signal f, and the value of the selected register is decremented by one. The value of the white level counter 16 is sent to the comparator 17 and compared with 0. If the value of the white level counter is equal to 0, the representative value of the white pixel is obtained, and the value e of the white register 11 is output as the representative value of the white pixel. If the value of the white level counter is not 0, the process returns to the beginning again, and the loop process starting from the value of the white register 11 being decoded by the decoder 12 is repeated. The initial value of the white level counter is a parameter, and when the pixels are arranged in the order of brightness, it represents which number of pixels from the brighter one is the representative brightness value of the white pixel. Generally, the value may be about several percent of the total number of pixels.

FIG. 3 is a block diagram of a representative pixel value detecting section for black pixels. In the figure, 21 is a black register, 22 is a decoder, and 23.
-1 to 23-n are registers, 24 is a comparator, 25 is an inverter, 26 is a black level counter, and 27 is a comparator. The data of the histogram created by the histogram creation unit is registered from register 0 (23-1)
It is stored in 255 (23-n). This register is the same as the histogram register of FIG. 2 or is another register to which data is copied. When using the same histogram register, it is necessary to prevent access to the histogram register at the same time. The initial value of the black register 21 is the minimum brightness value (0 in this case),
It is given before the loop processing described below is started.
The value after the loop processing is the representative value e ′ of the black pixel.

The beginning of the loop processing starts with the decoding of the value of the black register 21. The value of the black register 21 is decoded by the decoder 22, and the registers 23-1 to 2-2 in which the histogram value of the brightness value of the black register 21 is stored.
The control signal f'is determined so that any one of 3-n is selected. The histogram value g'read from the register selected only by the control signal f'is sent to the comparator 24 and compared with 0. Histogram value g '
Is equal to 0, the decrement signal h'is enabled by the comparator 24 and the value of the black register 21 is decremented by 1. If not 0, comparator 24 and inverter 2
5, the decrement signal i'becomes valid, and the histogram registers 23-1 to 23-n and the black level counter 26 are activated.
Sent to.

The decrement signal i'transmitted to the histogram register is effective only in the register selected by the control signal f ', and the value of the selected register is decreased by one. The value of the black level counter 26 is the comparator 2
Sent to 7 and compared to 0. If the value of the black level counter is equal to 0, it means that the representative value of the black pixel has been obtained, and the value e ′ of the black register 21 is output as the representative value of the black pixel. If the value of the black level counter is not 0, the process returns to the beginning again, and the loop process starting from the value of the black register 21 being decoded by the decoder 22 is repeated. The initial value of the black level counter is a parameter, and when the pixels are arranged in the order of brightness, it represents which number of pixels from the darker one is the representative brightness value of the black pixel. Generally, the value may be about several percent of the total number of pixels.

FIG. 4 is a diagram showing another embodiment (claim 2) of the image binarizing apparatus according to the present invention. In the figure, 31 is a block threshold detecting unit, 32 is a character / graphic block detecting unit, and 33 is a block. A non-character / graphic block threshold detection unit, 34 is a pixel threshold calculation unit, and 35 is a binarization unit. The pixel value j in the block is read from the original multivalued image. The pixel value j in the block is detected by the block threshold detection unit 31 as a binarization threshold. In order to detect the threshold value, for example, a histogram of the brightness values of the pixels in the block is taken, and the threshold value for binarization is obtained by the discriminant analysis method.

The pixel value j in the block is also sent to the character / graphic block detection unit 32 to detect whether the block contains a character, a graphic, or the like, or whether the block is plain. The method is to detect an edge by taking the absolute value of the difference between a certain pixel and the adjacent pixel, and if the maximum value of the difference absolute value in the block exceeds a certain threshold, the character or figure is blocked. It is determined to exist inside. Hereafter, for the sake of explanation, the block that is determined to have characters and figures will be
A block determined to have no character or graphic is called a non-character graphic block.

In the non-character / graphic block, the binarization threshold value k obtained from the block threshold detection unit 31 is ignored, and the non-character / graphic block threshold detection unit 33 obtains a new binarization threshold value. The method of calculating the threshold value in the non-character / graphic block threshold calculation unit 22 is, for example, whether or not there are character / graphic blocks in the upper, lower, left, and right blocks around the non-character / graphic block. If there is, the average value of the thresholds of the surrounding character / graphic blocks is calculated, and this is used as a new threshold value for binarization of the block. By repeating this series of processing, a new binarization threshold value can be obtained for all non-character graphic blocks. Since the non-character figure block threshold detection unit 33 has obtained the binarization threshold values n of all blocks, the pixel threshold value calculation unit 34 calculates the pixel-unit binarization threshold value from the block binarization threshold value.

FIG. 14 is a diagram for explaining an example of the calculation method of the pixel threshold value calculation unit in FIG. 1, in which blocks around the pixel 62 for which the binarization threshold value is desired are 50, 51, 52,
53, the central points of the blocks are 54, 55, 56 and 57, and the surrounding blocks 50, 51 and 5 are arranged so that the pixel 62 is contained in the area surrounded by the central points 54, 55, 56 and 57.
Choose 2,53.

The blocks are assumed to be rectangular. At this time, the center points 54, 55, 56, 5 of the position of the pixel 62 concerned
The ratio inside the rectangle 63 surrounded by 7 is a ₁ : b _{1 in} the vertical direction and c ₁ : d ₁ (a ₁ + b ₁ = 1, c ₁ + in the horizontal direction).
Let d ₁ = 1). The threshold of each block is made to correspond to the apex of the rectangle 63, and linear interpolation is performed according to the ratio to obtain the corresponding pixel 62.
The threshold value of is calculated. Set the threshold of the center point (vertex) to <54>, <
55>, <56>, <57>, the calculation formula is <54
> * B ₁ * d ₁ + <55> * b ₁ * c ₁ + <56> * a ₁ *
It becomes d ₁ + <57> * a ₁ * c ₁ . Pixel threshold calculation unit 34
The binarization threshold value p of each pixel obtained by
5 and compared with the pixel value j of the pixel concerned and binarized to form a binarized image.

FIG. 5 is a diagram showing an embodiment (Claim 3) of the image binarization device according to the present invention. In the figure, 44 is a threshold protruding block removing portion, and other portions which have the same function as FIG. Are given the same reference numerals. A threshold protruding block removing unit 44 is added to FIG. The threshold protruding block removing unit 4
4 detects, from among the threshold values obtained by the block threshold value detection unit 31, a value having a protruding value as compared with the surrounding blocks, and replaces the value with a value calculated from the threshold values of the surrounding blocks.

One method of detecting a block with a protruding threshold value is, for example, comparing the threshold value of the block with the threshold values of blocks in the neighborhood of 4 (or 8 neighborhoods), and determining whether the threshold of the block is the maximum value or the minimum value. Yes, and the maximum value (or minimum value, or the sum of the absolute values of the differences, etc.) of the difference between the threshold value of the block and the threshold values of the blocks in the surrounding 4 neighborhoods (or 8 neighborhoods) is a preset value. When it is larger, it is judged to be protruding. The neighborhood to compare is
In addition to the 4 or 8 neighborhoods, the above-described processing may be performed by comparing with the 4 neighborhoods of up, down, left and right, and then the same processing may be performed by comparing with the diagonal 4 neighborhoods.
In addition, one method of replacing the threshold value protruding from the surrounding blocks with a value calculated from the threshold values of the surrounding blocks is, for example, replacing with the average value of the threshold values of the blocks in the four neighboring areas (or eight neighboring areas). You may perform the process as mentioned above 2-3 times.

15 and 16 are flow charts showing the processing of the threshold protruding block removing unit. FIGS. 15 and 16 compare the threshold value with the blocks in the vicinity of four surroundings, and the threshold value of the block is the maximum or minimum, and the maximum absolute value of the difference from the threshold values of the blocks in the surrounding vicinity 4 is set in advance. 9 is a flowchart of a method of determining that the block is a threshold protrusion block when the value is larger than the threshold value, and setting the average value of the threshold values of blocks in the vicinity of four surroundings as the threshold value of the block.

In FIGS. 15 and 16, m is the number of blocks in the horizontal direction, and n is the number of blocks in the vertical direction.
i is an index indicating the position of the block in the horizontal direction, and takes an integer value from 0 to m-1. Further, j is an index indicating the position of the block in the vertical direction, and takes an integer value from 0 to n-1. th0 (i, j) is the threshold of the block at the position indicated by (i, j) obtained by the block threshold detection unit 31, and th1 (i, j) is
The threshold is the threshold of the block at the same position after the protruding block is removed.

In steps 60 to 67, it is judged whether the threshold value of the block is the maximum or minimum value by comparing with the threshold values of the blocks in the four surroundings. If the threshold value of the block is the maximum or minimum value, step 68 is executed. Through 71, it is determined whether the absolute value of the difference between the threshold value of the block and the threshold values of the blocks in the vicinity of the surrounding four is larger than a preset value. If the threshold value of the block is maximum or minimum and the absolute value of the difference from the threshold values of the blocks in the vicinity of the four surroundings is larger than a preset value, it is determined that the block is a threshold protruding block, and step 72 , The average value of the threshold values of the blocks in the vicinity of the four surroundings is calculated and used as the threshold value of the block.

Although not shown in the figure, when i = 0, the steps 60, 64 and 68 are skipped and j
= 0, the processing of steps 61, 65, and 69 is skipped, and when i = m-1, steps 62, 66, and 7 are skipped.
The process of 0 is skipped, and when j = n−1, the processes of steps 63, 67 and 71 are skipped. When i or j takes the values at both ends of the range as described above, the denominator in the calculation for obtaining the average of the threshold values of the neighboring blocks shown in step 72 is 2 or 3.

FIG. 6 is a diagram showing still another embodiment (claim 4) of the image binarizing apparatus according to the present invention. In the figure, 36 is a block threshold smoothing unit, and other than that, the same operation as in FIG. The same reference numerals are given to the parts that perform. A block threshold smoothing unit 36 is added to FIG. As one method for realizing the block threshold smoothing unit 36, a threshold value of a certain block up, down, left and right is added to the threshold value of the block, and the average value obtained by dividing by the number of blocks is set as a new threshold value of the block. . If the block is not the end, the number of blocks is 5.
If there are blocks at the end, the number of blocks is 2 or 4. By performing this process on all blocks, it becomes possible to smooth the threshold value between blocks.

FIG. 7 is a diagram showing still another embodiment (Claim 5) of the image binarizing apparatus according to the present invention. In the figure, 37 is an edge emphasizing section, and other functions are the same as those in FIG. The parts are given the same reference numerals. In addition to FIG. 6, the edge enhancement unit 37
Has been added. FIG. 7 is a diagram when edge enhancement is applied only to the original image referred to by the binarization unit. When edge enhancement is applied to the original image in advance, the image obtained by applying edge enhancement to the original image is changed to a new original image. If so, the configuration is the same as in FIG. 6 and is omitted. One method for realizing the edge enhancement unit 37 is to subtract a sum of pixel values of four pixels in the upper, lower, left, and right directions from a value obtained by multiplying the pixel value of the pixel by 5 to obtain a new pixel value of the pixel, and calculate the result. In the case of overflow or underflow, the maximum luminance value and the minimum luminance value are used as new pixel values for the pixel. This emphasizes the edges. The edge-enhanced image is processed by the binarization unit 3 as shown in FIG.
5, the block threshold detection unit 3
1, it may be input to the character / graphic block detection unit 32.

As described above, in the embodiment shown in FIGS. 4 to 7 (claims 2 to 5), the binarization threshold values of all blocks can be obtained by the block threshold value detection unit 31. For blank or blank areas, this threshold has no meaning. By the character / graphic block detection unit 32 and the non-character / graphic block threshold detection unit 33, appropriate threshold values can be obtained for all blocks, both blocks with and without characters and graphics. Even if there are shadows and brightness unevenness in the entire image, if the size of the block is sufficiently smaller than the size of the shadows and brightness unevenness, there will be no shadows or brightness unevenness only by looking inside the block. , This threshold is an appropriate threshold.

However, in an image in which illumination is reflected on a whiteboard or the like, the brightness unevenness due to the reflection may be smaller than the size of the block, and an appropriate threshold value can be obtained for a block that does not include reflection. On the other hand, an appropriate threshold value has not been obtained for the block including the reflection. In addition, a block including a glare may be determined by the character / graphic block detection unit 32 to be a character / graphic block due to the influence of the glare even on a plain portion where no characters or figures are present. Therefore, the non-character / graphic block threshold detection unit 33 may propagate an inappropriate threshold of a block including a glare to an adjacent non-character / graphic block. Such an inappropriate threshold has a characteristic that the value is more prominent than the thresholds of the surrounding blocks.
The threshold protruding block removing unit 44 can prevent such an inappropriate threshold value from being removed and propagating to an adjacent block.

When the threshold value obtained in each block and the pixel in the block are directly compared and binarized, the boundary between blocks having different threshold values becomes conspicuous. The pixel threshold calculation unit 34 calculates the threshold by interpolating the thresholds of the surrounding blocks for each pixel, and a smooth and natural binarized image can be obtained with no noticeable block boundaries. These enable binarization so that characters and the like can be recognized even in images with shadows and uneven brightness.

Further, since the threshold value obtained in each block is obtained only from the block concerned, it is an appropriate threshold value in each block, but it is a value protruding from the surroundings in the whole image, When the difference from the threshold is large, the binarization result of the entire image may become unstable. A more appropriate threshold value is obtained by smoothing the threshold value between blocks by the block threshold value smoothing unit 36.

When the number of pixels in the block is small,
The binarization threshold obtained in the block may not be optimal, and a thin line may cause the binarized image to become even thinner or crushed due to the non-optimal binarization threshold. By edge-enhancing the original multi-valued image and comparing the edge-enhanced pixel with a threshold value to perform binarization, such a detailed structure can be preserved even after binarization.

FIG. 8 is a diagram showing still another embodiment (Claim 6) of the image binarization apparatus according to the present invention, in which 38 is a block black and white representative pixel value detection unit and 39 is a character / graphic block detection. Part, 40 is a non-character figure block representative value calculation part, 41
Is a block threshold calculation unit, and the other parts that have the same functions as those in FIG. 4 are denoted by the same reference numerals.

The pixel value j in the block is read from the original multi-valued image. With respect to the pixel value in the block, the block monochrome representative pixel value detection unit 38 detects the representative value s of the white pixel and the representative value t of the black pixel. The representative value s of white pixels and the representative value t of black pixels obtained by the block monochrome representative pixel value detection unit 38.
The character / graphic block detection unit 39 detects whether the block includes a character, a graphic, or the like, or is solid. In this method, for example, the difference between the white pixel representative value s and the black pixel representative value t is calculated, and if the absolute value exceeds a certain threshold value, it is determined that a character or a figure exists in the block.

In the non-character graphic block, the representative pixel values s and t obtained from the block monochrome representative pixel value detection unit 38 are ignored, and the non-character graphic block representative value calculation unit 40 obtains a new representative pixel value. This processing is performed separately for the white pixel representative value and the black pixel representative value. The method of calculating the representative pixel value in the non-character / graphic block representative value calculation unit 40 is, for example, whether or not there are character / graphic blocks in the upper, lower, left and right blocks around the non-character / graphic block. If any,
The average value of the representative pixel values of the surrounding character / graphic blocks is calculated, and this is set as a new representative pixel value of the block. By repeating this series of processing, a new representative pixel value can be obtained for all non-character graphic blocks.

Non-character graphic block representative pixel value calculation unit 40
Since the white pixel representative pixel value v and the black pixel representative pixel value w of all blocks are obtained by the above, the block threshold value calculation unit 41
Then, the binarization threshold value x of each block is calculated. As a method of obtaining the binarization threshold value of the block, for example, the average of the white pixel representative value and the black pixel representative value may be calculated. Since the binarization thresholds of all blocks have been obtained, the pixel threshold calculation unit 34 obtains the binarization threshold for each pixel.

In FIG. 14, the blocks around the pixel 62 for which the binarization threshold is to be obtained are 50, 51, 52, 53, the center points of the blocks are 54, 55, 56, 57, and the center points 54, 55, The surrounding blocks 50, 51, 52, and 53 are selected so that the pixel 62 fits in the area surrounded by 56 and 57. The blocks are assumed to be rectangular. At this time, the ratio inside the rectangle 63 surrounded by the center points 54, 55, 56, 57 of the position of the pixel 62 is a ₁ : b _{1 in} the vertical direction and c ₁ : d ₁ (a ₁ + b _{1 in the} horizontal direction. = 1 and c ₁ + d ₁ = 1). The threshold of each block is made to correspond to the vertex of the rectangle 63,
Interpolation is performed according to the ratio to obtain the threshold value of the corresponding pixel 62. The threshold value of the center point (vertex) is set to <54>, <55>, <56>, <
57>, the calculation formula is <54> * b ₁ * d ₁ + <5
5> * b ₁ * c ₁ + <56> * a ₁ * d ₁ + <57> * a ₁
* C ₁ The binarization threshold value p of each pixel obtained by the pixel threshold value calculation unit 34 is sent to the binarization unit 35, is compared with the pixel value j of the pixel, and is binarized to form a binarized image.

FIG. 9 is a diagram showing an embodiment (Claim 7) of the image binarizing apparatus according to the present invention. In the figure, reference numeral 45 is a representative pixel value protruding block removing section, which has the same function as that of FIG. The same reference numerals are given to the parts to be performed. A representative pixel value protruding block removing unit 45 is added to FIG. The representative pixel value protruding block removing unit 45 protrudes the value of the representative value s of the white pixels and the representative value t of the black pixels obtained by the block monochrome representative pixel value detecting unit 38 as compared with the surrounding blocks. An object is detected and its value is replaced with a value calculated from the white pixel representative value or the black pixel representative value of the surrounding blocks.

One method of detecting a block in which the black-and-white representative pixel value is prominent is, for example, that the white (black) pixel representative value of the block is changed to the white (black) pixel representative value of blocks in the neighborhood of four (or eight) neighborhoods. And the white (black) pixel representative value of the block is the maximum value or the minimum value, and the white (black) pixel representative value of the block and the surrounding 4 neighborhoods (or 8 neighborhoods)
Maximum absolute value of the difference from the white (black) pixel representative value of the block (or the minimum value, or the sum of absolute differences)
Is larger than a preset value, it is determined to be protruding. In addition to 4 neighborhoods or 8 neighborhoods, the neighborhoods to be compared are compared with the 4 neighborhoods in the upper, lower, left, and right directions, and the above-described processing is performed, and then the similar processings are performed in comparison with the blocks in the 4 diagonal neighborhoods. It may be one. Further, one method of replacing the white (black) pixel representative value protruding from the surrounding block with the value calculated from the white (black) pixel representative value of the surrounding block is, for example, a block in the neighborhood of 4 (or 8) in the neighborhood. Replace with the average value of the white (black) pixel representative values. You may perform the process as mentioned above 2-3 times.

A flow chart of the concrete processing is shown in FIG.
16 and FIG. 16 is exactly the same as the flowchart of the threshold protruding block removing unit shown in FIG.
th0 (i, j) is represented by the block monochrome representative pixel value detection unit 3
The black-and-white representative pixel value of the block at the position indicated by (i, j), th1 (i, j), obtained in step 8, is replaced with the black-and-white representative pixel value of the block at the same position after the removal of the representative pixel value protruding block. Just do it.

FIG. 10 is a diagram showing still another embodiment (claim 8) of the image binarizing apparatus according to the present invention, in which 42 is a block black-and-white representative pixel value selecting section, and FIG. Portions having the same function are designated by the same reference numerals. A block monochrome representative pixel value selection unit 42 is added to FIG. The block black-and-white representative pixel value selection unit 42 selects which of the white pixel representative value s and the black pixel representative value t of the block determined to be a non-character / graphic block by the character / graphic block detection unit 39 is reliable.

As a selection method, for example, if a character / graphic block exists in a block around the non-character / graphic block, the difference between the white pixel representative value of the character / graphic block and the white pixel representative value of the non-character / graphic block is calculated. The difference between the black pixel representative value of the character / graphic block and the black pixel representative value of the non-character / graphic block may be compared, and the smaller difference may be regarded as more reliable.
For the pixel representative value of the non-character figure block determined to be unreliable, the average of the pixel representative values of the surrounding character figure blocks is used as a new value. When the pixel representative values for both white and black are obtained, the block is set as a character / graphic block, and all non-character / graphic blocks become character / graphic blocks.

FIG. 11 is a diagram showing still another embodiment (claim 9) of the image binarization apparatus according to the present invention, in which 43 is a block black-and-white representative pixel value smoothing unit, and FIG. The same reference numerals are attached to the parts having the same functions as. Figure 10
In addition, a block monochrome representative pixel value smoothing unit 43 is added. As one method of realizing the block black-and-white representative pixel value smoothing unit 43, the representative pixel values of the upper, lower, left, and right blocks of a certain block are added to the representative pixel values of the block,
The value averaged by dividing by the number of blocks is set as a new representative pixel value of the block. The number of blocks is 5 for blocks other than the ends, and the number of blocks is 3 or 4 for blocks at the ends. By performing this process for all blocks, it becomes possible to smooth representative pixel values between blocks.

FIG. 12 is a view showing still another embodiment (claim 10) of the image binarizing apparatus according to the present invention, and the reference numerals in the drawing are the same as those in FIGS. 7 and 10. A block threshold smoothing unit 36 is added to FIG. As a method of realizing the block threshold smoothing unit 36, a threshold value of blocks above and below and to the left and right of a block and a threshold value of the block are added, and a value averaged by dividing by the number of blocks is a new threshold value of the block. And The number of blocks is 5 for blocks other than the ends, and the number of blocks is 3 or 4 for blocks at the ends. By performing this process on all blocks, it becomes possible to smooth the threshold value between blocks.

FIG. 13 is a view showing still another embodiment (claim 11) of the image binarization apparatus according to the present invention, and the reference numerals in the drawing are the same as those in FIGS. 7 and 11. In addition to FIG. 11, an edge emphasis section 37 is added. FIG. 13 is a diagram when edge enhancement is applied only to the original image referred to by the binarization unit. When edge enhancement is applied to the original image in advance, the image obtained by applying edge enhancement to the original image is changed to a new original image. If so, the configuration is the same as that in FIG. 12, and is omitted. One method of realizing the edge enhancement unit 37 is to set the pixel value of the pixel to 5
A value obtained by subtracting the sum of the pixel values of four pixels in the upper, lower, left, and right directions from the multiplied value is set as a new pixel value of the pixel. When the calculation result overflows or underflows, the maximum brightness value and the minimum brightness value are set as the new pixel value of the pixel. This is a method of setting different pixel values. This emphasizes the edges. The edge-enhanced image is shown in FIG.
In addition to being used only by the binarizing unit 35 as in 3, the input may be made to the block monochrome representative pixel value detecting unit 38.

As described above, in the embodiments shown in FIGS. 8 to 13 (claims 6 to 11), the block black-and-white representative pixel value detection unit 38 obtains the representative values of the white pixels and the black pixels of all the blocks. Even if there are shadows and brightness unevenness in the entire image, if the size of the block is sufficiently smaller than the size of the shadows and brightness unevenness, there will be no shadows or brightness unevenness only by looking inside the block. , The representative pixel value is an appropriate representative pixel value.

However, in an image in which the illumination is reflected on the wide board, the brightness unevenness due to the reflection may be smaller than the size of the block. In a block that does not include the reflection, an appropriate representative pixel is included. While the value is obtained, an appropriate representative pixel value is not obtained in the block including the reflection. In addition, a block including glare may be determined by the character / graphic block detection unit 39 to be a character / figure block due to the effect of glare, even in a plain portion without characters or figures.

Therefore, the non-character graphic block representative value calculation unit 40 may propagate an inappropriate representative value of a block including a glare to an adjacent non-character graphic block. Since such an inappropriate representative pixel value has a feature that the value is more prominent than the representative pixel values of the surrounding blocks, the representative pixel value protruding block removing unit 45
It is possible to prevent such inappropriate representative pixel values from being removed and propagating to adjacent blocks. For blocks with letters and figures, both black and white representative values are reliable,
In a blank or plain area that does not include characters or figures, either the white or black representative value selected by the block black-and-white representative pixel value selection unit 42 becomes reliable.

When the threshold value is obtained for each block, the threshold value of a blank or plain portion that does not include characters or figures has no meaning, and the non-character figure block representative value calculation unit 40 must interpolate from the surroundings. However, when a plain area occupies a large area, it may not be able to sufficiently follow changes such as shadows and uneven brightness. When calculating the black and white representative pixel value, either the white or black representative value can be used, and that value should reflect changes such as shadows and brightness unevenness, so the solid part Even in the case of occupying a large area, even if the other value is interpolated by the non-character figure block representative value calculation unit 40, it is possible to sufficiently follow changes such as shadows and uneven brightness.

By the character / graphic block detection unit 39, the non-character / graphic block representative value calculation unit 40, and the block black-and-white representative pixel value selection unit 42, appropriate threshold values are set for all blocks, both with and without characters and graphics. can get. The binarization threshold value of each block is calculated by the block threshold value calculation unit from the white pixel representative value and the black pixel representative value of the block. If the threshold value obtained in each block and the pixel in the block are directly compared and binarized, the boundary between blocks having different threshold values becomes conspicuous. The pixel threshold calculation unit 34 calculates the threshold by interpolating the thresholds of the surrounding blocks for each pixel, and a smooth and natural binarized image can be obtained with no noticeable block boundaries. These enable binarization so that characters and the like can be recognized even in images with shadows and uneven brightness.

Further, since the representative value obtained in each block is obtained only from the block concerned, it is an appropriate representative value in each block, but when viewed from the whole image, it becomes a value protruding from the surroundings or is adjacent. When the difference from the representative value of the block to be processed is large, the binarization result of the entire image may become unstable. By the block monochrome representative pixel value smoothing unit 43,
By smoothing the representative value between blocks, a more appropriate representative value can be obtained. Similarly, the threshold value obtained in each block is not only the representative value, but the threshold value obtained in each block is obtained only from the block, so it is an appropriate threshold value in each block, but when viewed from the entire image, it becomes a value protruding from the surroundings, When the difference between the thresholds of adjacent blocks is large, the binarization result of the entire image may become unstable. The block threshold smoothing unit 36 smoothes the thresholds between blocks to obtain a more appropriate threshold.

When the number of pixels in the block is small,
The binarization threshold obtained in the block may not be optimum, and the thin line may cause the binarized image to become thinner or crushed due to the non-optimal binarization threshold. By edge-enhancing the original multi-valued image and comparing the edge-enhanced pixel with a threshold value to perform binarization, such a detailed structure can be preserved even after binarization.

[0059]

As is apparent from the above description, the present invention has the following effects. (1) When the pixels are arranged in the order of brightness using a histogram, the brightness value of which number from the lightest one is the white pixel representative value, and the value of which brightness value from the darkest one is the black pixel representative value are binarized. The binarization threshold value can be obtained at high speed by setting the threshold value to the average of the white pixel representative value and the black pixel representative value. Further, particularly in a character image, the binarization threshold can be determined in the valley portion between the peaks of black pixels and the peaks of white pixels in the histogram, so that the binarized image also gives good results. In the case of an image with only thin lines, only one of the black pixel peaks or the white pixel peaks produces a clear peak, and the other may be widely and thinly dispersed, but the value of the portion far from the dispersed portion By taking as the representative value, the average of the white pixel representative value and the black pixel representative value is less likely to be overlaid on the mountain, so that it becomes an appropriate binarization threshold value. Further, since the threshold for binarization is directly obtained from the histogram without using the evaluation function, the threshold can be obtained at high speed. (2) The binarization thresholds of all blocks can be obtained by the block threshold detection means, but this threshold has no meaning for a blank sheet or a plain portion that does not include characters or figures. Therefore, the character / graphic block detection means and the non-character / graphic block threshold calculation means can obtain appropriate threshold values for all blocks, whether or not there are blocks with characters and graphics. Even if there are shadows and brightness irregularities in the entire image, if the size of the block is sufficiently smaller than the size of the shadows and brightness irregularities, there will be shadows and brightness irregularities only in the block. Therefore, this threshold is an appropriate threshold. (3) If there is a small amount of brightness and unevenness compared to the size of the block due to the reflection of illumination, etc., an appropriate block threshold value or white / black representative pixel value cannot be obtained. Since the threshold value and the white / black representative pixel value are projected as compared with the surrounding blocks, they can be removed by the threshold projecting block removing means or the representative pixel value protruding block removing means, even though it is a white background portion. It is also possible to prevent an obstacle that the reflection part is binarized to white and the ground part is binarized to black due to the influence of the reflection. (4) When the threshold value obtained in each block and the pixel in the block are directly compared and binarized, the boundary between blocks having different threshold values becomes conspicuous. Therefore, the image threshold calculation unit interpolates the thresholds of the surrounding blocks for each pixel to calculate the thresholds, and a smooth and natural binarized image can be obtained with no noticeable block boundaries. These enable binarization so that characters and the like can be recognized even in images with shadows and uneven brightness. (5) Further, the threshold value obtained in each block is an appropriate threshold value in each block because it is obtained only from the relevant block, but it is a value protruding from the surroundings in the whole image, When the difference from the threshold is large, the binarization result of the entire image may become unstable. Therefore, a more appropriate threshold value can be obtained by smoothing the threshold value between blocks by the block threshold value smoothing means. (6) Further, when the number of pixels in the block is small, the binarization threshold obtained in the block may not be optimal, and a thin line may make the binarized image thinner due to the non-optimal binarization threshold. May be crushed. Therefore, the original multi-valued image is edge-emphasized, and the edge-enhanced pixel is compared with a threshold value to be binarized, whereby such a detailed structure can be preserved even after binarization.

[Brief description of drawings]

FIG. 1 is a configuration diagram for explaining an embodiment (claim 1) of an image binarization apparatus according to the present invention.

FIG. 2 is a configuration diagram of a representative pixel value detection unit (white pixel) in FIG.

FIG. 3 is a configuration diagram of a representative pixel value detection unit (black pixel) in FIG.

FIG. 4 is a diagram showing another embodiment (claim 2) of the image binarizing apparatus according to the present invention.

FIG. 5 is a diagram showing still another embodiment (Claim 3) of the image binarization apparatus according to the present invention.

FIG. 6 is a diagram showing still another embodiment (claim 4) of the image binarizing apparatus according to the present invention.

FIG. 7 is a diagram showing still another embodiment (Claim 5) of the image binarizing apparatus according to the present invention.

FIG. 8 is a diagram showing still another embodiment (Claim 6) of the image binarization apparatus according to the present invention.

FIG. 9 is a diagram showing still another embodiment (Claim 7) of the image binarization apparatus according to the present invention.

FIG. 10 is a diagram showing still another embodiment (Claim 8) of the image binarization apparatus according to the present invention.

FIG. 11 is a diagram showing still another embodiment (Claim 9) of the image binarizing apparatus according to the present invention.

FIG. 12 is a diagram showing still another embodiment (Claim 10) of the image binarization apparatus according to the present invention.

FIG. 13 is a diagram showing still another embodiment (Claim 11) of the image binarizing apparatus according to the present invention.

FIG. 14 is a configuration diagram of an image threshold calculation unit in the present invention.

FIG. 15 is a flowchart (part 1) showing the processing of the threshold protrusion block removal unit in FIG. 5;

16 is a flowchart (No. 2) showing the processing of the threshold protruding block removal unit in FIG.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Signal processing part, 2 ... A / D conversion part, 3 ... Histogram creation part, 4 ... Representative pixel value detection part, 5 ... Threshold value determination part, 31
... block threshold detection unit, 32 ... character / graphic block detection unit, 33 ... non-character / graphic block threshold detection unit, 34 ... pixel threshold calculation unit, 35 ... binarization unit, 36 ... block threshold smoothing unit, 37 ... edge enhancement Part 38 ... Block black-and-white representative pixel value detection unit 39 ... Character / graphic block detection unit 40 ... Non-character / graphic block representative value calculation unit 41 ... Block threshold value calculation unit 42 ... Block black / white representative pixel value selection unit 43 ... Block monochrome representative pixel value smoothing unit, 44 ... Threshold protruding block removing unit, 45 ... Representative pixel value protruding block removing unit

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification number Internal reference number FI Technical indication location G06K 9/38 A (72) Inventor Sadahiko Higami 22-22 Nagaikecho, Abeno-ku, Osaka-shi, Osaka Prefecture Co., Ltd. (72) Inventor Hiroyoshi Toda 22-22 Nagaike-cho, Abeno-ku, Osaka-shi, Osaka Prefecture Sharp Corporation (72) Inventor Toshifumi Horikawa 22-22 Nagaike-cho, Abeno-ku, Osaka City, Osaka Prefecture

Claims (11)

[Claims] 1. Histogram creating means for finding a histogram of brightness values from a multi-valued image, and white pixel values in the image from the histogram found by the histogram creating means,
Representative pixel value detection means for obtaining a value representative of the black pixel value,
A white pixel representative value obtained by the representative pixel value detecting means,
An image binarization device, comprising: a threshold value determining unit that obtains a threshold value for binarization from a black pixel representative value. 2. A block threshold value detecting means for obtaining a threshold value for binarization for each block of a multi-valued image divided into blocks, and a character / graphic block detecting means for detecting whether a character or a figure is included in the block. , A non-character / graphic block threshold calculation means for calculating a binarization threshold of a block determined by the character / graphic block detection means that does not include a character or graphic, from the thresholds of surrounding blocks, and A pixel threshold value calculating means for converting the binarization threshold value into a threshold value for each pixel, and 1 obtained by the pixel threshold value calculating means.
An image binarization device, comprising: a binarization unit that binarizes an original multi-valued image using a binarization threshold value for each pixel. 3. A threshold protruding block for replacing a threshold value of a block having a value that is higher than that of a surrounding block among the threshold values obtained by the block threshold value detecting means, with a value calculated from the threshold values of the surrounding blocks. The image binarization apparatus according to claim 2, further comprising a removing unit. 4. A block threshold smoothing means for reducing a difference between a threshold of a certain block and a threshold of a surrounding block with respect to a threshold for each block obtained by the non-character figure block threshold detecting means. The image binarization apparatus according to claim 2 or 3, characterized in that. 5. An edge enhancing means for applying edge enhancement to a multi-valued image to be compared with a threshold value by the binarizing means, or an edge enhancing means for applying edge enhancement to an original multi-valued image in advance. The image binarization device according to claim 2, 3 or 4. 6. A block black and white representative pixel value detecting means for obtaining a representative value of white pixels and a representative value of black pixels for each block of a multi-valued image divided into blocks, and whether or not a character or figure is included in the block. And a non-character graphic block representative for calculating a representative pixel value of a block determined by the character / graphic block detection unit that does not include a character or a graphic from the representative pixel values of surrounding blocks. Value calculation means, block threshold calculation means for calculating the binarization threshold of the block from the representative pixel values of the white pixels and black pixels calculated by the non-character figure block representative value calculation means, and two calculated for each block. An original multi-valued image is binarized by using a pixel threshold value calculation means for converting the threshold value for binarization into a threshold value for each pixel and a binarization threshold value for each pixel obtained by the pixel threshold value calculation means. An image binarization apparatus comprising: a binarization unit. 7. Among the representative values of the white pixels and the black pixels obtained by the block black and white representative pixel value detecting means, the representative pixel value of the block whose value is higher than that of the surrounding block is set to the surrounding block. 7. The image binarization apparatus according to claim 6, further comprising: representative pixel value protruding block removing means for replacing the representative pixel value with a value calculated from the representative pixel value. 8. A block black-and-white representative for selecting which of a white pixel representative value and a black pixel representative value of the block is reliable for a block determined by the character / graphic block detecting means to include no characters or figures. It is characterized by comprising pixel value selecting means and non-character graphic block representative value calculating means for calculating the representative pixel values not selected by the block monochrome representative pixel value selecting means from the representative pixel values of surrounding blocks. The image binarization device according to claim 6. 9. A difference between a representative value of a certain block and a representative value of surrounding blocks is reduced with respect to the representative value of the white pixel and the black pixel for each block obtained by the non-character figure block representative value calculating means. 9. A block black-and-white representative pixel value smoothing means for performing the above is provided.
The image binarization apparatus described. 10. The block threshold smoothing unit is configured to reduce a difference between a threshold of a certain block and a threshold of a surrounding block with respect to the binarization threshold for each block obtained by the block threshold calculation unit. 9. The image binarization device according to claim 6, 7 or 8. 11. An edge enhancing means for applying edge enhancement to a multi-valued image to be compared with a threshold value by the binarizing means, or an edge enhancing means for applying edge enhancement to an original multi-valued image in advance. 6. The method according to claim 6, wherein
0. The image binarization apparatus according to any one of items.