JP2000316074A - Scanner device and noise removal method - Google Patents

Abstract

(57) [Problem] In a handy scanner having a structure in which it is difficult to provide a shield portion due to its structure, noise is likely to be superimposed on a scanned image under the influence of a commercial power supply or the like. A handy scanner capable of obtaining a high-quality image by temporarily taking an image into a memory and performing a noise removal process is realized. SOLUTION: An image for one screen is once taken into a memory,
An estimation error is determined from a difference between the weighted average of the pixel in the sub-scanning direction and the pixel. The error of the line including the pixel is determined by averaging the estimated error in the main scanning direction, and the noise component is removed by subtracting the error from the scanned value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a small-sized and low-cost handy scanner, which effectively removes noise of a read image and reduces the noise, and a method of removing the noise.

[0002]

2. Description of the Related Art In a scanner device for taking in an image using a so-called line sensor in which sensors are arranged in a straight line, each pixel is read by switching each read pixel of the line sensor in the main scanning direction, and the sensor section is manually operated. A so-called handy scanner which reads pixels in the sub-scanning direction by detecting the amount of movement of the sensor unit with a movement amount detecting means such as a rotary encoder connected to a paper pressing roller via a gear or the like.

In this system, the apparatus can be made more compact than a flatbed system in which a line sensor moves inside the scanner main body, and a sheet feed mechanism provided with a mechanism for transporting paper because the scanner main body does not require a wide flat surface. There is a feature that it is possible to read a thick original such as a book which is impossible with the method.

In the handy system, since the movement in the sub-scanning direction is performed manually and the sensor is pressed against the document manually, rollers and pads for pressing the document from the back surface as in other methods are not required. The feature is that the device can be configured to be very compact. In order to further utilize this feature, a contact sensor that forms a one-to-one virtual image directly on the sensor using a selfoc lens array can be used. This eliminates the need for a reduction optical system, and can achieve further compactness.

[0005] By adopting such a configuration, a very small scanner device can be obtained, but a high-sensitivity sensor element is exposed to the space through a material having no effect of blocking electromagnetic waves, such as glass or a lens. Therefore, the structure is easily affected by electromagnetic waves in the space. Since there is no housing on the back side of the document, a shield including the document cannot be realized.

Of the electromagnetic waves in the space, those having a frequency substantially equal to or less than the sub-scanning cycle of the sensor, for example, noise originating from a commercial power supply, have a remarkably adverse effect on the image quality. May cause a specific sub-scanning component to become brighter or darker, or an image to appear colored when it occurs in a specific color.

[0007]

In a scanner device using this contact type sensor, a commercial power supply (50 Hz AC) is used.
Or, when used in a space where low-frequency electromagnetic noise generated at 60 Hz) exists, a line sensor included in the scanner picks up the low-frequency electromagnetic noise and mixes stripe noise into an image.

In a handy scanner, the following two points are cited as factors which are susceptible to this noise.

First, since there is no scanner housing on the back of the document, a metal shield plate cannot be provided.
By providing a metal plate connected to the ground on the back surface of the document, the sensor is surrounded by metal, so that low-frequency noise can be reduced. Of course, the surface of the manuscript
Since the front of the sensor is in the direction in which light rays from the original image enter, a shield that blocks light cannot be provided.

In a scanner having a reduction optical system,
By bending the optical path with a mirror, the normal direction of the sensor can be directed to a direction different from that of the original, so that a metal shield plate can be provided on the surface facing the sensor.

Second, because of the handy operation, the reading speed of the original in the sub-scanning direction is uneven, and the period of the noise is not constant. Therefore, a specific and constant frequency component (for example, 50 Hz or 60 H
Noise cannot be removed by the method of removing z).

The present invention realizes noise reduction even in a handy sensor in which such a small sensor is a close contact type.

[0013]

SUMMARY OF THE INVENTION In order to solve this problem, the present invention detects a noise level due to a commercial power supply or the like by reading in advance a uniform color or plain paper and examining pixel fluctuations in that case. I do. Next, an image for one screen is accumulated from the line sensor, a noise component is estimated for this image, and an image with less noise is obtained by removing the noise component.

First, for a pixel from which noise is to be removed,
An average of the read values of the pixels for the n pixels before and after in the sub-scanning direction is taken to obtain a temporary estimated value of the pixel. Since a general image has continuity in color tone, noise components are canceled by taking an average. The difference between the provisional estimated value and the actually read value is determined, and the provisional error of the pixel is determined. By averaging the tentative error for one line in the main scanning direction, an average error value in that line is obtained.

In this case, when a component exceeding the level of noise obtained by reading the above-described uniform paper is detected, the change is not due to the noise but to a change in the color data of the original image. It is determined that there is, and is excluded from the calculation for obtaining the temporary error shown above. As a result, an average error value in a line including the pixel is obtained.

The average error value of one main scanning line, which is the expected value of the error obtained by the above operation, is uniformly subtracted from all the pixels of the main scanning line as the error of the main scanning line. This makes it possible to reduce noise components due to the commercial power supply, and to obtain an image with less noise.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The invention according to claim 1 of the present invention is directed to a line sensor having pixels arranged on a line in the main scanning direction and read data in the main scanning direction and the sub-scanning direction, and In a scanner device that reads a pixel in the sub-scanning direction by moving, a tentative estimation value of the pixel is obtained by weighting and averaging the values of a plurality of upper and lower pixels in the sub-scanning direction for a target pixel, and the value of the pixel and the tentative estimation are obtained. The difference between the tentative estimated value and the tentative estimated value exceeds a predetermined value. A noise removal method for a scanner device, wherein noise of the pixel of interest is removed by subtracting the average error value from the value of the pixel of interest. In this case, a handy scanner that can obtain an output image with less noise even when the scanner sensor receives low-frequency noise generated from a commercial power supply (50 Hz or 60 Hz AC) or the like.

According to a sixth aspect of the present invention, in the case where a pixel value of a plurality of upper and lower pixels in the sub-scanning direction is weighted and averaged with respect to a target pixel in one main scanning line,
2. The weight of pixels at both ends in the sub-scanning direction, i.e., a pixel just in the range to be weighted and a pixel at a position out of the range to be weighted and averaged is less than 1, and the weight of other pixels is 1. The scanner device of the above, when obtaining the optimal estimated value using the upper and lower n pixels of the pixel of interest in the sub-scanning direction, reduces the effect of the end value of the upper and lower n pixels, the influence of the value of the read image A scanner device that realizes less noise removal is realized.

According to an eighth aspect of the present invention, in the case where a pixel value of a plurality of upper and lower pixels in the sub-scanning direction is weighted and averaged for a pixel of interest in one main scanning line,
2. The scanner device according to claim 1, wherein a weight of the pixel is set to 1, and a window function is used to decrease the weight as the distance from the pixel is increased. By reducing the weight as the distance increases, it is possible to realize a scanner that can effectively remove noise from an image having a more gradual change in color tone such as a natural image.

FIG. 1 shows an embodiment of the present invention.
This will be described with reference to FIG.

(Embodiment 1) FIG. 1 is a block diagram of a scanner portion according to an embodiment of the present invention. Reference numeral 1 denotes a line sensor in which sensors for converting the light intensities of RGB into electric signals are arranged in a line. Reference numeral 2 denotes a sensor driving unit which reads out pixels in the main scanning direction by sequentially reading out the intensity of light detected by each of the line sensors 1 from the sensors. Reference numeral 3 denotes an A / D converter, which converts the voltage value of the output signal of the line sensor into a value that can be digitally processed. Reference numeral 4 denotes a shading correction unit, which varies the sensitivity of the sensor element, the illumination of the original,
Variations due to the optical system are normalized based on a previously captured white image, and variations due to these factors are removed. Reference numeral 10 denotes a roller, which improves the slip of the handy scanner when the sensor is manually moved, and regulates the moving direction so that the line sensor 1 moves in parallel.
Reference numeral 11 denotes a gear, which transmits the rotation of the roller 10 to the movement amount detection unit 12 in the sub-scanning direction. Reference numeral 5 denotes a memory transfer unit that transfers image data to the memory 6 based on the output of the shading correction unit and the result of the movement amount detection unit 12. Reference numeral 7 denotes a processing unit which removes noise by processing an image once stored in the memory 6. Reference numeral 8 denotes an output interface unit which reads out image data from which noise has been removed from the memory 6 and outputs it.

These elements are not limited to the case where all of them are incorporated in the handy scanner. Similar effects can be expected even if the memory 6 and the processing unit 7 are configured in an information processing apparatus connected to the handy scanner.

The operation of removing noise in the handy scanner having these configurations will be described in detail.

First, a plain or uniform color sheet is traced with a handy scanner. Originally, the data obtained as a result shows a uniform value. However, in practice, the value varies due to the influence of noise such as a commercial power supply. By calculating the difference between the maximum value and the minimum value, a noise range can be obtained.

First, in the case of a color image, the above processing
The same process is performed separately for each color (eg, R, G, B). This is because it is difficult to use for noise estimation because the average value is completely different for each color. Unless otherwise specified, the same processing is performed for each of the RGB colors.

Next, the original to be read is read by a scanner. The intensity of light detected by the line sensor 1 is converted into an electric signal, which is sequentially read out in the main scanning direction by the sensor driving unit 2 and sent to the A / D converter 3. The A / D converter 3 quantizes and digitizes the analog electric signal. Generally, the electric signal is from 256 gradations to 65 for each color of RGB.
The quantization is performed on the K gradation. The signal quantized by the A / D converter 3 is stored in the memory 6 for one line after a shading correction unit performs a correction process for variations in sensitivity of the sensor element, variations in illumination of the original, and variations due to the optical system.

When the line sensor 1 is moved by hand, the rollers 10 are regulated so as to rotate by the same amount at both ends, so that the rollers 10 move smoothly in the parallel direction. This movement amount is transmitted to the movement amount detection unit 12 via the gear 11, and the movement amount detection unit 12 adjusts the movement amount for the movement amount of 25.4 mm.
Pulses from 0 to 600 are generated to output the movement amount. When the movement amount detector 12 generates a pulse, the line sensor 1 can read the image in the sub-scanning direction by reading the image in the same manner as described above.

Images read by these operations are all stored in the memory 6 in the main scanning direction and the sub-scanning direction. At this point, since the data stored in the memory 6 includes noise due to the influence of the commercial power supply or the like, the noise is removed using the following method.

The range in the sub-scanning direction (the numerical value represented by n in the above description) referred to when calculating the average value for each pixel is determined by the balance between the image and the information processing power of the apparatus. Since it is necessary to cancel that each line receives different non-periodic positive and negative noises, a certain size is required. However, if this value is increased to 678..., The amount of calculation increases in proportion, and the change of the image itself is averaged, resulting in a flat image.

Conversely, if n is reduced, the ability to average noise (cancel positive and negative noise) also decreases. Generally, noise decreases in inverse proportion to the square root of N, where N is the average number of pixels. Therefore, N is determined in consideration of image evaluation and processing time. In this example, N = about 10.

The weight of the weighted average is the same regardless of the distance from the pixel of interest, and is set at the end, that is, the position just entering the weighted average range and the position outside the range when the next weighted average is performed. Only a certain pixel has a small weight of about 0.5. The weight distribution is set so as to suppress the fluctuation of the tentative estimated value due to the noise received by the line that is out of the range of the weighted average or that enters the line.

[0032]

[Table 1]

In Table 1, assuming that the pixel of interest is P6, a value of P6 that does not include noise is first determined. The upper and lower 11 pixels (upper 5 pixels, lower 5 pixels, and a total of 11 pixels) in the sub-scanning direction of P6 are averaged to obtain a temporary estimated value P6 ′ of the pixel value at P6. The difference between this and the value of P6 is eP6
= P6-P6 ′. Similarly, the error, eP, for all pixels (A6, B6,.
A, ePB. . . Find ePZ.

By averaging the estimation error in the sub-scanning direction with respect to the main scanning direction, an average error value which is an error of a line including the pixel of interest (P6) is estimated. External noise, such as a commercial power supply, which causes the above-mentioned error, cannot exceed a certain level. If the difference between the maximum value and the minimum value obtained when the plain paper is read is determined, it is determined that the image level itself has changed rather than the influence of noise, and is excluded from the calculation of the average value.

Since the average error value obtained in this manner is an optimum estimate of noise received from the outside of the line, the average error value is subtracted from the value read by the line sensor for each image value of the line. However, if the result exceeds the maximum value and the minimum value that the image can take, the image is saturated with the maximum value and the minimum value.

In the above description, the reference range is limited to the upper and lower 5 pixels. However, the reference range is limited to the upper and lower 5 pixels depending on the cycle of one main scanning line of the image, the frequency relationship of the commercial power supply, the noise source and the resolution. is not.

As described above, by making the weights of the pixels located at both ends of the reference range smaller than the weights of the other pixels, the estimation value and the estimation error are reduced due to the influence of the pixels distant from the pixel of interest. be able to. That is, a number of pixel values are averaged to obtain a stable average value. However, when averaging 11 pixels,
If the reference range is shifted by one, one pixel goes out of the reference range and at the same time one pixel enters the reference range. If the signs of these errors are the same, there is no effect, but if they are different, the influence of the pixel whose average is relatively far from the pixel of interest is large (2
/ 11 pixels). To prevent this, the weights of the pixels at both ends of the reference range are set to １ ／, and the influence is reduced even if the sign of the error of the incoming and outgoing pixels differs between positive and negative. (Estimate 1/10 = (1/2 + 1/2) / (9 + 1/2)
+1/2)) Of course, if the error of the pixel one inward from the end of the reference range moves in the same way as the error of the end pixel, the above effect is lost, but at the same time, the four pixels move +++-. The probability is small and the effect on image quality can be small.

(Embodiment 2) Since the basic circuit configuration and the operation except for noise removal are the same as those in Embodiment 1, the description is omitted, and the noise removal operation of the handy scanner will be described in detail.

First, a plain or uniform color sheet is traced with a handy scanner. Originally, the data obtained as a result shows a uniform value. However, in practice, the value varies due to the influence of noise such as a commercial power supply. By calculating the difference between the maximum value and the minimum value for each of the RGB colors, a noise range can be obtained. In the following description, the same processing is performed for each color, so that the same processing is performed for all three colors of RGB unless otherwise specified.

Next, the original to be read is read by the scanner. The intensity of light detected by the line sensor 1 is converted into an electric signal, which is sequentially read out in the main scanning direction by the sensor driving unit 2 and sent to the A / D converter 3. The A / D converter 3 quantizes and digitizes the analog electric signal. Generally, the electric signal is from 256 gradations to 65 for each color of RGB.
The quantization is performed on the K gradation. The signal quantized by the A / D converter 3 is stored in the memory 5 for one line after the shading correction unit performs a correction process for variations in sensitivity of the sensor element, variations in illumination of the original, and variations due to the optical system.

When the line sensor 1 is moved by hand, the roller 10 is regulated to rotate by the same amount at both ends, so that the roller 10 moves smoothly in the parallel direction. This movement amount is transmitted to the movement amount detection unit 12 via the gear 11, and the movement amount detection unit 12 adjusts the movement amount for the movement amount of 25.4 mm.
Pulses from 0 to 600 are generated to output the movement amount. When the movement amount detector 12 generates a pulse, the line sensor 1 can read the image in the sub-scanning direction by reading the image in the same manner as described above.

Images read by these operations are all stored in the memory 6 in the main scanning direction and the sub-scanning direction. At this point, since the data stored in the memory 6 includes noise due to the influence of the commercial power supply or the like, a method of removing the noise will be described.

Weighted averaging is performed using a window function in which the weight decreases as the distance from the pixel of interest increases. Since the tone of a general natural image changes in a gradual manner in most cases, a pixel adjacent to a pixel of interest is a color that is close in many cases. In addition, by reducing the weight as the distance increases, the influence of a change in color of the image to be read itself can be reduced.

In Table 1, assuming that the pixel of interest is P6, first, a value including no noise of P6 is obtained. A tentative estimated value P6 'of the pixel value at P6 is obtained from a total of 11 pixels of the upper and lower 5 pixels in the sub-scanning direction of P6.

P6 '= (1 / 5P1 + 2 / 5P2 + 3 /
5P3 + 4 / 5P4 + P5 + P6 + P7 + 4 / 5P8 +
3 / 5P9 + 2 / 5P10 + ／ P11) / 7 The difference between this and the value of P6 is eP6 = P6-P6 ′.
Similarly, all pixels in the main scanning direction (A6, B in the figure)
6. . . Z6), error, ePA, ePB. . . eP
Find Z.

The error of the line including the pixel of interest (P6) is estimated by averaging the estimation error in the sub-scanning direction with respect to the main scanning direction. External noise, such as a commercial power supply, which causes the above-mentioned error, cannot exceed a certain level. Those exceeding the difference between the maximum value and the minimum value obtained when the plain paper is read are judged to have changed the image level, and are excluded from the calculation of the average value.

The average value obtained in this manner is subtracted from the line because the line is the optimum estimated value of the noise received from the outside. However, if the result exceeds the maximum value and the minimum value that the image can take, the image is saturated with the maximum value and the minimum value.

In the above description, the reference range is limited to the upper and lower five pixels. However, according to the relationship between the cycle of one main scanning line of an image and the frequency of external AC noise, it is better to use other parameters.

Further, by making the weights of the pixels located at both ends of the reference range smaller than the weights of the other pixels, the estimation value and the estimation error are influenced by the pixels distant from the pixel of interest. Can be reduced.

[0050]

As described above, according to the present invention, even in a handy scanner in which a shield plate for AC noise cannot be provided and the reading cycle is not constant,
In a handy scanner device that cannot remove noise by simply removing specific frequency components by extracting noise components due to the influence of commercial power from the image once read by the scanner and subtracting from the read image data to correct it, And a good image with few images can be obtained.

The present invention realizes an extremely excellent handy scanner capable of obtaining a good image with such a simple configuration.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a scanner device according to an embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Line sensor 2 Sensor drive part 3 A / D converter 4 Shading correction part 5 Memory transfer part 6 Memory 7 Processing part 8 Output interface 10 Roller 11 Gear 12 Movement amount detection part

Claims (8)

[Claims] A line sensor having pixels arranged in a line in the main scanning direction with respect to read data in a main scanning direction and a sub-scanning direction, and a scanner device for reading pixels in the sub-scanning direction by moving the line sensor. A weighted average of the values of the upper and lower pixels in the sub-scanning direction with respect to the pixel of interest to obtain a tentative estimation value of the pixel, obtain a difference between the value of the pixel and the tentative estimation value, An average error value is obtained by averaging one line in the main scanning except for those exceeding the value. A noise removing method for a scanner device, wherein noise of the target pixel is removed by subtracting the average error value from the value of the target pixel. 2. A scanner device for reading data in a main scanning direction and a sub-scanning direction having pixels arranged on a line in the main scanning direction, and a scanner device for reading pixels in the sub-scanning direction by moving the line sensor. A weighted average of the values of the upper and lower pixels in the sub-scanning direction with respect to the pixel of interest to obtain a tentative estimation value of the pixel, obtain a difference between the value of the pixel and the tentative estimation value, An average error value is obtained by averaging one line in the main scanning except for those exceeding the value. A scanner device, wherein noise of the target pixel is removed by subtracting the average error value from the value of the target pixel. 3. A line sensor having pixels arranged on a line in the main scanning direction and reading data in the main scanning direction and a sub-scanning direction, and a scanner device for reading pixels in the sub-scanning direction by moving the line sensor. 2. The method according to claim 1, wherein the difference between the maximum value and the minimum value of the values read from the plain or uniform scanned object is a predetermined value. 4. A paper scanning roller comprising: a paper pressing roller; and a moving amount detecting unit for obtaining a moving amount from a rotation angle of the roller, wherein a sub-scanning is performed based on a value detected by the moving amount detecting unit. 3. The scanner device according to claim 2, wherein an amount of movement in the direction is obtained. 5. A weighted average of pixel values of a plurality of upper and lower pixels in a sub-scanning direction with respect to a pixel of interest in one line of a main scanning line, which is located at both ends in the sub-scanning direction, that is, within a weighted average range. 2. The noise removal method for a scanner device according to claim 1, wherein the weight of the pixel which is out of the range of the next weighted average with the pixel just after is set to less than 1, and the weight of other pixels is set to 1. 6. When weighting and averaging pixel values of a plurality of upper and lower pixels in a sub-scanning direction with respect to a pixel of interest in one line of a main scanning direction, both ends in the sub-scanning direction, that is, within a weighted average range. 2. The scanner device according to claim 1, wherein the weight of a pixel which is out of the range of the weighted average and the pixel just after is set to less than 1, and the weight of other pixels is set to 1. 7. When weighting the pixel values of a plurality of upper and lower pixels in the sub-scanning direction with respect to a pixel of interest in one line of main scanning of interest, the weight of the pixel is set to 1, and the weight increases as the distance from the pixel increases. 2. The method according to claim 1, wherein a window function is used to reduce the noise. 8. A weighted average of pixel values of a plurality of upper and lower pixels in a sub-scanning direction with respect to a pixel of interest in one line of a main scanning line of interest. 2. The scanner device according to claim 1, wherein a window function is used to reduce the following.