JP2019092101A - Image processing apparatus, image processing method, and program - Google Patents

Abstract

To suppress the occurrence of moire in an image after halftone processing even when an image before the halftone processing includes a periodic pattern in an image processing apparatus for generating a periodic dot pattern in the halftone processing.SOLUTION: An image processing apparatus includes: filter processing means that generates a band pass image by performing filter processing of extracting a component of a predetermined spatial frequency range on the basis of an input image; determination means that determines whether there is a correlation between a target pixel and peripheral pixels of the target pixel on the basis of the band pass image; smoothing means that performs smoothing on the basis of the input image according to the correlation determination result output by the determination means; and halftone processing means that generates a dot pattern on the basis of the output image of the smoothing means.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a technology for removing a periodic pattern that causes moire generation.

  BACKGROUND In recent years, MFPs (Multi-Function-Printers) having a plurality of functions such as scanners, printers, copiers, and fax machines have become widespread. Image forming apparatuses for realizing a printer function in an MFP are roughly classified into an electrophotographic method and an inkjet method. In electrophotography, the color material formation on the recording medium is unstable compared to the inkjet method. Therefore, in general, halftone processing often generates a periodic dot pattern instead of a random dot pattern. . This achieves low graininess, that is, low noise image output.

  The periodic dot pattern is generated, for example, by dithering using an AM screen in a threshold matrix. A halftone screen (dot screen) or a line screen is known as an AM screen. On the other hand, a random dot pattern is generated by, for example, dither processing using an FM screen in a threshold matrix, or error diffusion processing. As an FM screen, a blue noise mask and a green noise mask are known.

  In an electrophotographic image forming apparatus in which color material formation on a recording medium is unstable, low graininess, that is, low noise image output can be realized by generating a periodic dot pattern in halftone processing. However, if the image before halftoning contains a periodic pattern, it interferes with the period (period of the threshold) of the AM screen used in halftoning, and a low frequency pattern appears in the image after halftoning. (Moire) occurs. As a result, the image quality of the printed matter is greatly degraded.

  The best case in which the image before halftone processing includes a periodic pattern is a copy of a print on the MFP. Since periodic dot patterns such as halftone dot patterns and line patterns are widely used not only for electrophotographic printers, but also for halftone reproduction in general printing for newspapers and magazines, they can be copied. Many printed items that contain a periodic pattern. Therefore, when an AM screen is used in halftone processing at the time of copying a printed matter, moiré occurs in the copied output in many cases.

  In addition, even in the case of printing a digital document, moiré may occur in the printed matter. For example, when converting vector data such as a presentation document into a raster image, translucent objects included in the vector data are often rasterized as a periodic pattern. Since a raster image containing this periodic pattern is used in halftone processing after undergoing color conversion processing, gamma correction processing, etc., when an AM screen is used in halftone processing, moiré occurs in the printed output. It will

  Although the moiré shown above can be reduced by generating a random dot pattern in halftone processing, when the image forming apparatus is an electrophotographic method, the formation of color materials on the recording medium is unstable, so graininess can be reduced. That is, the noise gets worse. Therefore, in the electrophotographic system, it is difficult to achieve both reduction of moiré and low graininess (low noise).

  As a technique capable of solving such a problem, Patent Document 1 selectively smoothes only periodic patterns such as halftone dots while maintaining the shape of an edge in a stage prior to performing halftone processing. Technology is disclosed. Specifically, the entire surface of the input image is smoothed to remove the periodic patterns such as halftone dots, and the input image before smoothing is alpha-blended with different weights for each place. The weight for each place is determined by local feature quantities such as dot degree and edge degree generated by the image area determination means, and in the area with high dot degree, the blend ratio of the smoothed image becomes large, and the edge degree In the high region, the blending ratio of the image before smoothing is large. As a result, since it is possible to remove the periodic pattern that interferes with the AM screen at the stage before performing the halftone process, it is possible to suppress the occurrence of moiré even if the AM screen is used in the halftone process. In addition, since the shape of the edge is maintained, a high-quality printed output can be obtained as compared with the case where the entire surface of the input image is smoothed.

  Patent Document 2 discloses a technique for removing noise while maintaining the shape of an edge by selectively smoothing an input image. Specifically, first, filtering is applied to the input image to increase the power of the first frequency band including the signal component, and to decrease the power of the second frequency band including the noise component. This generates a reference image with an improved SN ratio compared to the input image. Then, based on the correlation determination result in the reference image, the input image is selectively smoothed. Thereby, the noise component can be suitably removed while holding the signal component such as the edge or the texture.

JP 2008-011268 A JP, 2013-115697, A

  However, in the technique described in Patent Document 1, since smoothing is performed across the edge, if the blend ratio of the smoothed image is adjusted so as to completely remove the periodic pattern, the edge may be blurred. There were many cases. On the other hand, when the blend ratio of the smoothed image is reduced to maintain the shape of the edge, there are many cases where the periodic pattern remains near the edge. If the local feature quantity calculation method used to generate the blend ratio is improved to solve this problem, the circuit scale tends to increase and the mounting cost tends to increase. Furthermore, even if the method of calculating the feature value is improved by multiplying the expensive mounting cost, it is difficult to blend an image that has been smoothed across the edge at a suitable ratio, so a sufficient effect can not be obtained. The

  On the other hand, since the technique described in Patent Document 2 can perform smoothing without straddling edges by determining correlation, the technique described in Patent Document 1 can be used as a technique for selectively smoothing an input image. There is an advantage in comparison. In addition, since a circuit for calculating a local feature amount used to generate the blend ratio is unnecessary, the increase in the mounting cost as described above can be suppressed.

  However, when generating the reference image used for the correlation determination, the frequency characteristics of the signal component and the noise component are considered, but the frequency characteristics of the periodic pattern interfering with the AM screen are not considered. Therefore, when the technology described in Patent Document 2 is applied to an image before halftone processing, a periodic pattern that interferes with the AM screen remains, or conversely, the input image is overblown and a high quality print is obtained. There is a problem that no output can be obtained.

  Therefore, in view of such circumstances, the present invention provides an image processing apparatus for generating a periodic dot pattern in halftone processing, wherein even if an image before halftone processing includes a periodic pattern, the halftone processing is performed. The purpose is to suppress the occurrence of moiré in a later image.

The present invention performs filter processing for extracting a component in a predetermined spatial frequency range based on an input image, thereby generating a band pass image, a target pixel based on the band pass image, and the target pixel. Judgment means for judging whether there is a correlation with peripheral pixels of the pixel, smoothing means for carrying out smoothing based on the input image according to the correlation judgment result outputted by the judgment means, and output image of the smoothing means And an halftone processing unit that generates a dot pattern based on the above.

  According to the present invention, in an image processing apparatus that generates a periodic dot pattern in halftone processing, even if the image before halftone processing includes a periodic pattern, moire in the image after halftone processing The occurrence can be suppressed.

Block diagram showing the configuration of the image processing apparatus in the first embodiment Block diagram showing the configuration of the image processing unit in the first embodiment Block diagram showing the configuration of the periodic pattern removal circuit in the first embodiment The figure which shows the view territory in execution example 1 Filter coefficient and amplitude characteristic of the filter processing unit in the first embodiment Block diagram showing the configuration of the periodic pattern removal circuit in the second embodiment Block diagram showing the configuration of the periodic pattern removal circuit in the third embodiment Block diagram showing the configuration of the periodic pattern removal circuit in the fourth embodiment

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. The following embodiments are specific examples of the present invention, and therefore, various technically preferable limitations are added. However, the embodiments are not limited to the following embodiments or other specific methods as long as the idea of the present invention is followed.

Example 1
<About the configuration of the image processing apparatus>
The image processing apparatus according to the first embodiment will be described below with reference to FIG. FIG. 1 is a block diagram showing an example of the overall configuration of the image processing apparatus according to this embodiment. As illustrated, the image processing apparatus includes a CPU circuit unit 110, an image reading unit 120 that scans an original 100 and reads an image, an image processing unit 130, and a printer unit 140. The image reading unit 120 includes a lens 122, a CCD sensor 124, and an analog signal processing unit (A / D conversion unit) 126. The image of the original 100 formed on the CCD sensor 124 through the lens 122 is converted by the CCD sensor 124 into analog electric signals of R (Red), G (Green), and B (Blue).

  The image information converted into the analog signal is input to the analog signal processing unit 126, and after correction etc. is performed for each of the R, G and B colors, it is subjected to analog / digital conversion (A / D conversion). The digitized full color signal (hereinafter referred to as a digital image) is input to the image processing unit 130. The image processing unit 130 subjects the digital image to input correction processing to be described later, periodic pattern removal processing, color space conversion processing, density correction processing, and halftone processing, and the digital image after these processings is Output to the printer unit 140.

  The printer unit 140 is an image forming apparatus that employs an electrophotographic method, an inkjet method, or the like, and forms an image on a recording medium based on an input digital image. The CPU circuit unit 110 further includes a CPU 112 for arithmetic control, a ROM 114 for storing fixed data and programs, a RAM 116 for temporarily storing data and loading a program, and an external storage device 118. The CPU circuit unit 110 (of the CPU 112) controls the image reading unit 120, the image processing unit 130, and the printer unit 140 in an integrated manner, and executes the processing in this embodiment described below. The external storage device 118 is a storage medium such as a hard disk storing parameters and programs used by the image processing apparatus in the present embodiment. Data, programs and the like developed in the RAM 116 may be loaded from the external storage device 118 instead of the ROM 114.

<About the configuration of the image processing unit>
Hereinafter, the image processing unit 130 in the present embodiment will be described in detail with reference to FIG. FIG. 2 is a block diagram showing an example of the circuit configuration of the image processing unit 130 in this embodiment.

  As illustrated, the image processing unit 130 outputs the input interface 210, the input correction circuit 220, the periodic pattern removal circuit 230, the color space conversion circuit 240, the density correction circuit 250, the halftone processing circuit 260, and And an interface 270. When a digital image is input from the analog signal processing unit 126 to the image processing unit 130 via the bus 205, processing described below is performed in each of the above-described components of the image processing unit 130.

  The digital image 215 is input to the input correction circuit 220 via the input interface 210. The digital image 215 is composed of R, G and B luminance signals. The input correction circuit 220 performs processing on the digital image 215 to correct variations in the characteristics of sensors that read the document 100 and variations in the light distribution characteristics of the lamp for document illumination, and the period of the processed digital image 225 is It is output to the pattern removal circuit 230.

  A digital image 225 composed of the luminance signals of R, G and B output from the input correction circuit 220 is input to the periodic pattern removal circuit 230. The periodic pattern removal circuit 230 performs periodic pattern removal processing on the digital image 225, and outputs the digital image 235 after the periodic pattern removal processing to the color space conversion circuit 240. The periodic pattern removal process will be described in detail later (see FIG. 3).

  A digital image 235 composed of the luminance signals of R, G, and B output from the periodic pattern removal circuit 230 is input to the color space conversion circuit 240. The color space conversion circuit 240 is configured by the digital image 235 composed of luminance signals of R, G and B by the density signals of C (Cyan), M (Magenta), Y (Yellow) and K (Black). The digital image 245 is converted and output to the density correction circuit 250.

  A digital image 245 composed of C, M, Y and K density signals output from the color space conversion circuit 240 is input to the density correction circuit 250. The density correction circuit 250 performs density correction processing on the digital image 245, and outputs the digital image 255 after the density correction processing to the halftone processing circuit 260. The density correction circuit 250 performs density correction processing in advance in consideration of the characteristics of the halftone processing circuit 260 so that a density change does not occur when the subsequent halftone processing circuit 260 performs binarization or multi-value conversion. Keep it.

  A digital image 255 composed of C, M, Y, and K density signals output from the density correction circuit 250 is input to the halftone processing circuit 260. The halftone processing circuit 260 subjects the digital image 255 to halftone processing, specifically, halftone screen processing with a predetermined number of lines. As a result, the digital image 255 in which each pixel has a multi-value pixel value is converted into a digital image of halftone expression, that is, a digital image 265 in which each pixel has a binary or multi-value pixel value. Then, the halftone processing circuit 260 outputs the digital image 265 to the printer unit 140 via the output interface 270 and the bus 275.

  In halftone processing (halftone screen processing) in the halftone processing circuit 260, a periodic dot pattern is generated. As a result, particularly when the electrophotographic method is adopted for the printer unit 140, it is possible to realize low graininess, that is, low noise image output, as compared with the case where random dot patterns are generated. The periodic dot pattern is generated by a known method such as dithering using an AM screen in a threshold matrix, for example. In the case of binary dither processing, when the pixel value before halftone processing exceeds the value in the threshold matrix, a value indicating dot ON is output, while the pixel value is less than or equal to the value in the threshold matrix, A value indicating dot OFF is output. As the AM screen, a dot screen or a line screen is used. The halftone processing circuit 260 only needs to generate a periodic dot pattern based on the digital image 255 composed of C, M, Y, and K density signals, and the method used is not limited to dither processing.

<About periodic pattern removal processing>
The periodic pattern removal process performed by the periodic pattern removal circuit 230 according to this embodiment will be described below with reference to FIG. The periodic pattern removal circuit 230 executes the same processing for each color component, and FIG. 3 is a block diagram showing an example of the periodic pattern removal circuit 230 that performs processing for one color component. As illustrated, this circuit includes a filter processing unit 301, a correlation determination unit 302, and a smoothing unit 303.

  In FIG. 3, i (x, y) indicates one color component (one of R, G, and B) of the digital image 225 output from the input correction circuit 220. At this time, x and y indicate positions in the image. The resolution is 600 dpi. Hereinafter, i (x, y) is referred to as an input image.

  The filter processing unit 301 generates a band pass image b (x, y) by performing a convolution operation on the input image i (x, y) and the filter coefficient k (x, y). The filter coefficient k (x, y) is a design parameter, and changing the value can change the frequency characteristic of the filter. Filter coefficients having desired frequency characteristics are generated by a known method such as a window function method. When the window function method is used, when u and v are spatial frequencies in the x and y directions and F (u, v) is the target frequency characteristic, F (u, v) is inverse Fourier transformed to obtain a window function By limiting the filter size, filter coefficients having desired characteristics can be obtained. The features of the filter used in the filter processing unit 301 will be described in detail later (see FIG. 5).

  The correlation determination unit 302 determines whether or not there is a correlation between the pixel of interest and each pixel around the pixel of interest (referred to as a peripheral pixel) in the region of interest. The smoothing unit 303 performs smoothing based on the correlation determination result c (x, y) output from the correlation determination unit 302 using only peripheral pixels that have a correlation with the pixel of interest in the region of interest. The values of the pixels used for smoothing are the pixel values of the input image i (x, y). Then, the value obtained by the smoothing is derived as the value of the target pixel. o (x, y) is an image with values derived in this way. Hereinafter, o (x, y) is called an output image. The output image o (x, y) is output to the color space conversion circuit 240 as one color component of the digital image 235.

  Hereinafter, processing in the correlation determination unit 302 and the smoothing unit 303 will be described in detail with reference to FIG. FIG. 4 shows a three-pixel square region adjacent to the pixel of interest as an example of the region of interest.

In each of the focus areas shown in FIG. 4, the focus pixel is the central pixel, and in the focus areas 401 to 403, the subscript to the focus pixel is “11”. The focused area 401 is an area in the input image i (x, y), and i _ij in the focused area 401 is a pixel value of the input image i (x, y). An attention area 402 is an area in the band pass image b (x, y) output by the filter processing unit 301, and b _ij in the attention area 402 is a pixel value of the band pass image b (x, y). An attention area 403 is an area in the correlation determination result c (x, y) output from the correlation determination unit 302, and c _ij in the attention area 403 is a pixel value of the correlation determination result c (x, y).

The band pass image b (x, y) generated by the filter processing unit 301 is used for the correlation determination, and c _ij is calculated according to equation (1). In Equation (1), the pixel with c _ij = 1 is a pixel determined to have a correlation (correlated) with the target pixel. On the other hand, the pixel with c _ij = 0 is a pixel determined to have no correlation (no correlation) with the target pixel. That is, the correlation determination unit 302 obtains a difference absolute value between the pixel of interest of the band pass image and its peripheral pixels in the region of interest, and the difference absolute value is equal to a predetermined threshold (referred to as a threshold for correlation determination) Th. It is determined whether it exceeds (in other words, whether it is less than or equal to a predetermined threshold). As a specific example of the correlation determination, reference numeral 404 in FIG. 4 is a diagram showing an example of a focused region in the band pass image, and reference 405 indicates a case where the correlation determination threshold Th is 5 for the focused region 404. It is a figure which shows a correlation determination result.

The smoothing unit 303 averages the pixel values i _ij in the input image for one or more pixels determined to be correlated (c _ij = 1) in the region of interest. For example, if the correlation determination result 405, the value i _avg obtained by averaging is _{"(i 00+ i 10+ i 20+ i} 01+ i 11+ i 21) ÷ 6 ". Then, the target pixel value i ₁₁ is replaced by the i _avg. An output image o (x, y) is obtained by performing the above processing on the entire image.

  In the above description, although the size of the focus area in the correlation determination and the smoothing is set to three pixels in a square, the size may be other than that. Also, the size of the region of interest may be adaptively switched depending on the mode of halftone processing. Specifically, the lower the spatial frequency of the dot pattern generated by the halftone processing, the lower the frequency component contributing to the moire. Generally, a filter of large size is required to remove low frequency components. Therefore, the removal performance of the low frequency component can be secured by increasing the size of the above-mentioned focused area as the spatial frequency of the dot pattern generated by the halftone processing is lower. On the other hand, by making the size of the region of interest smaller as the spatial frequency of the dot pattern generated by the halftone processing becomes higher, it is possible to prevent the low frequency component from being overwhelmed.

<About the feature of the filter used in the filter processing unit>
Hereinafter, features of the filter coefficient k (x, y) used in the filter processing unit 301 will be described with reference to FIG.

  The band-pass image b (x, y) is desirably an image in which frequency components that interfere with the periodic dot pattern generated by the halftone processing circuit 260 are suppressed, and further, low-frequency components near the direct current are suppressed. In order to obtain such a band pass image b (x, y), in the present embodiment, frequency components in a predetermined spatial frequency range are extracted, while frequency components in other ranges are extracted, as described below. Use a band pass filter to suppress. For example, filter coefficients of 11 × 11 size as indicated by reference numeral 501 in FIG. 5 are used. The frequency characteristic of the filter coefficient 501 is substantially isotropic, and for example, the amplitude characteristic in the x direction is indicated by reference numeral 502.

  The periodic dot pattern generated by the halftone processing circuit 260 depends on the mode of halftone processing, but generally the spatial frequency is often higher than 6 cycles / mm. Moire is particularly noticeable when the frequency component of the input image is close to the spatial frequency of the periodic dot pattern generated by halftone processing. Therefore, as indicated by reference numeral 502, the filter coefficient 501 is made to have a characteristic of suppressing a component having a frequency higher than that near 6 cycles / mm.

  Further, the range in which the gradient is generated by the frequency component becomes wider as the low frequency. Therefore, the low frequency component near the direct current causes a gradient not only at the edge portion but also at the edge peripheral portion, which adversely affects the aforementioned correlation determination and smoothing. Specifically, as shown in equation (1), pixels with large gradients are considered to be pixels that cross the edge and are excluded from the smoothing targets. However, if a low frequency component near the direct current is included, a large gradient occurs not only in the edge part but also in the edge peripheral part, so that the smoothing in the edge peripheral part becomes insufficient. In order to solve this problem, if the value of the correlation determination threshold Th is increased to increase the number of pixels to be smoothed, not only the edge periphery but also the edge is blurred, and the image quality is significantly degraded. For this reason, the filter coefficient 501 is set to suppress low frequency components near DC as indicated by reference numeral 502.

  On the other hand, in order to retain the shapes of characters and edges, the filter coefficient 501 is set to pass frequency components other than the above. Components having frequencies higher than 6 cycles / mm of characters and edges are removed by the filter having the characteristic indicated by reference numeral 502. However, since characters and edges have many components near 3 cycles / mm, it is possible to make a difference by passing frequency components near 3 cycles / mm using a filter with the characteristic shown by reference numeral 502, but characters And can maintain the shape of the edge. By using such a band pass image for correlation determination, it is possible to suitably remove only frequency components that cause moiré without smoothing characters and edges.

  In the band-pass image, components higher in frequency than 6 cycles / mm are suppressed, and periodic patterns that cause moire generation are removed, but characters and edges are blurred. However, since the shapes of characters and edges are retained, if this band-pass image is used for correlation determination, characters and edges are not smoothed, and in the image after smoothing, blurring of characters and edges is prevented. be able to. On the other hand, since the periodic pattern that causes moire generation is removed in the band pass image, if this band pass image is used for correlation determination, the periodic pattern that causes moire generation can be smoothed. As described above, at the time of the band pass image, not only the periodic pattern is removed but also the characters and the edges are blurred, but in this embodiment, this band pass image is used for correlation determination and input Smooth the image. This is a major feature of the present embodiment in that a smoothed image in which only the periodic pattern is removed is obtained without blurring of characters and edges.

  In the above example, the filter processing unit 301 uses the 11 × 11 size filter, but the filter size may be other than 11 × 11.

  In addition, the filter characteristic is not limited to the characteristic indicated by reference numeral 502. For example, when the spatial frequency of the periodic dot pattern generated by the halftone processing circuit 260 is higher than 8 cycles / mm, the characteristic shown by reference numeral 502 is not passed, and the vicinity of 6 cycles / mm is allowed to pass. We adopt the characteristic of suppressing high frequency components. Thereby, the shapes of characters and edges can be maintained.

  Also, the filter coefficient used in the filter processing unit 301 may be switched depending on the mode of halftone processing. Specifically, the lower the spatial frequency of the dot pattern generated by the halftone processing, the lower the frequency component contributing to the moire. Therefore, as the spatial frequency of the dot pattern generated by halftone processing is lower, a filter with a lower cutoff frequency is used. This makes it possible to reliably remove frequency components that cause moiré. On the other hand, as the spatial frequency of the dot pattern generated by the halftone processing is higher, blurring of characters and edges can be prevented by using a filter with a high cutoff frequency.

  Further, the upper limit of the spatial frequency range of the component to be extracted by the filter processing unit 301 may be set according to the number of screen lines of the halftone screen processing executed by the halftone processing circuit 260. That is, as the number of screen lines increases, the upper limit of the spatial frequency range of the component extracted by the filter processing unit 301 is set higher.

<About the effect of this embodiment>
According to the present embodiment, in the image processing apparatus that generates a periodic dot pattern in halftone processing, even if the image before halftone processing includes a periodic pattern, moire in the image after halftone processing Can be suppressed. By generating a periodic dot pattern in halftone processing, in particular, when an electrophotographic method is adopted for the printer unit 140, an image with low graininess, that is, low noise, as compared to the case where a random dot pattern is generated. Output can be realized.

  Further, according to the present embodiment, since the smoothing can be performed without crossing the edge by determining the correlation, a high-quality smoothed image can be obtained as compared to the technique described in Patent Document 1. In addition, in the present embodiment, a circuit for calculating the local feature amount used for generating the blend ratio is unnecessary, so that the increase in the mounting cost can be suppressed as compared with the technique described in Patent Document 1.

  Further, in the present embodiment, unlike the technique described in Patent Document 2, when generating a reference image (band pass image) used for correlation determination, as shown by reference numeral 502 in FIG. Use a filter that takes into account the frequency characteristics of the typical pattern. This filter has a characteristic of removing a periodic pattern that causes moiré. Therefore, in the technique described in Patent Document 2, there is a problem that a periodic pattern that interferes with the AM screen remains, or conversely, the input image is excessively blurred, but according to the present embodiment, high image quality is obtained. A printed output is obtained.

  Moreover, in the technique described in Patent Document 2, when generating a reference image used for correlation determination, the power of the first frequency band including the signal component is increased. In a general image, the low frequency component near the direct current contains a large amount of signal components. Therefore, according to the technique described in Patent Document 2, the low frequency component near the direct current is amplified. As described above, low frequency components near the direct current cause gradients not only at the edge but also at the edge periphery, which adversely affects the correlation determination and smoothing. However, according to the configuration of the present embodiment, as shown by reference numeral 502, a filter that suppresses low frequency components near the direct current is used, so that smoothing at the edge periphery is insufficient or vice versa. It can prevent you from doing too much.

  Moreover, in the technique described in Patent Document 2, when generating a reference image used for correlation determination, the power of the first frequency band including the signal component is increased. In general, amplifying the power of frequency components expands the dynamic range. Therefore, in order to reliably represent the gradation value of the reference image, it is necessary to increase the number of bits of the reference image, which is disadvantageous from the viewpoint of the implementation cost. However, according to the configuration of the present embodiment, as indicated by reference numeral 502, the dynamic range of the reference image is not expanded because the maximum value of the amplitude characteristic of the filter for generating the reference image is set to 1 or less. Therefore, it is not necessary to increase the number of bits for expressing the gradation value of the reference image, and the mounting cost can be suppressed.

Example 2
In the first embodiment, when a band pass image used for correlation determination is generated, characters of small size are smoothed, and in correlation determination, it is not distinguished from periodic patterns causing moiré occurrence, and as a result, smooth. In the conversion unit 303, there is a risk of being deceived in the same manner as the periodic pattern.

  By the way, in the input image, when only a periodic pattern causing moiré occurrence is present, the pixel values of the band pass image become uniform over a wide range. On the other hand, if a small character exists, the pixel values of the band pass image become uniform near the center of the character, and it may not be distinguishable from the periodic pattern at the time of correlation determination. The pixel value often changes near the boundary of the pixel. Therefore, by evaluating the change of the pixel value of the band pass image over the range of a plurality of pixels, it is possible to distinguish between characters of small size and periodic patterns.

  In this embodiment, the change of the pixel value of the band pass image is evaluated in the 3 × 3 size target area shown in FIG. 4, that is, the amount of change is derived, and the threshold value for correlation determination Th according to the derived amount of change. Vary by location. Specifically, in a place where the amount of change is large, it is determined that a character or an edge is included in the region of interest, and the correlation determination threshold Th is set small to improve the retention performance of the shape of the character or the edge. Conversely, in places where the amount of change is small, it is determined that the area of interest does not include anything other than the periodic pattern, and the correlation determination threshold Th is set large to enhance the periodic pattern removal performance. In the following, differences from the above-described embodiment will be mainly described, and the description of the same contents as the above-described embodiment will be appropriately omitted.

<About periodic pattern removal processing>
Hereinafter, the periodic pattern removal process performed by the periodic pattern removal circuit 230 in the present embodiment will be described using FIG. Also in the present embodiment, in the periodic pattern removal circuit 230, the same processing is executed for each color component as in the first embodiment. FIG. 6 is a block diagram showing an example of a circuit for performing processing on one color component (any one of R, G, and B) of the periodic pattern removal circuit 230 in the present embodiment.

  As illustrated, the periodic pattern removal circuit 230 in the present embodiment includes a filter processing unit 601, a correlation determination unit 602, and a smoothing unit 603. These are similar to the filter processing unit 301, the correlation determination unit 302, and the smoothing unit 303 in the first embodiment. However, the periodic pattern removal circuit 230 in the present embodiment is different from the first embodiment in that a threshold value setting unit 604 is provided.

  The threshold setting unit 604 derives the change amount E of the pixel value of the band pass image b (x, y) from the values of a plurality of pixels forming the focused area 402. For example, by calculating the change amount E according to the equation (2), the circuit can be shared with the calculation for correlation determination shown in the equation (1), and an increase in mounting cost can be suppressed. The method of deriving the change amount E is not limited to the method using the equation (2), and any method may be used as long as the amount of change in the pixel value of the bandpass image within the range of a plurality of pixels can be evaluated. Absent.

  Then, when the change amount E is larger than the predetermined threshold value Th_E, it is determined that a character or an edge is included in the target area, and the correlation determination unit 602 sets a small value Th_1 as the correlation determination threshold Th used. In order to improve the retention of character and edge shapes. On the other hand, when the change amount E is smaller than the predetermined threshold value Th_E, it is determined that the area of interest does not include anything other than the periodic pattern, and the correlation determination threshold Th is set to a value Th_2 larger than Th_1. Improve pattern removal performance. When the change amount E is equal to the predetermined threshold value Th_E, one of Th_1 and Th_2 may be set as the correlation determination threshold value Th. As described above, in the present embodiment, the correlation determination threshold Th is set so as to suppress the smoothing in the smoothing unit 603 as the change amount E becomes larger.

<About the effect of this embodiment>
According to the present embodiment, it is possible to preferably remove only the periodic pattern that is a cause of the occurrence of moiré without turning small characters.

[Example 3]
In this embodiment, the input image is emphasized by adding the band pass image multiplied by the gain to the input image. The other configuration is the same as that of the first embodiment.

<About periodic pattern removal processing>
The periodic pattern removal process executed by the periodic pattern removal circuit 230 in the present embodiment will be described below using FIG. Also in the present embodiment, in the periodic pattern removal circuit 230, the same processing is executed for each color component as in the first embodiment. FIG. 7 is a block diagram showing an example of a circuit for performing processing on one color component (any one of R, G and B) of the periodic pattern removal circuit 230 in the present embodiment.

  As illustrated, the periodic pattern removal circuit 230 in the present embodiment includes a filter processing unit 701, a correlation determination unit 702, and a smoothing unit 703. These are similar to the filter processing unit 301, the correlation determination unit 302, and the smoothing unit 303 in the first embodiment. However, the periodic pattern removal circuit 230 in the present embodiment is different from the first embodiment in that the multiplication unit 704 and the addition unit 705 are provided.

  The multiplication unit 704 multiplies the band pass image b (x, y) by a predetermined gain coefficient G, and outputs an emphasis amount d (x, y) obtained thereby to the addition unit 705. The addition unit 705 adds the amount of enhancement d (x, y) to the input image i (x, y), and obtains an image (hereinafter referred to as an enhanced image) e (x, y) in which the sharpness is obtained. The signal is output to the smoothing unit 703. The smoothing unit 303 performs smoothing processing on the emphasized image e (x, y), and outputs an output image o (x, y) obtained thereby to the color space conversion circuit 240.

  The filter processing unit 701 uses a filter coefficient having substantially zero amplitude characteristics at direct current (0 cycle / mm) as indicated by reference numeral 502 in FIG. 5. The reason for this is that if a filter coefficient whose amplitude characteristic at direct current (0 cycle / mm) is not zero is used, the average value of the emphasis amount d (x, y) does not become substantially zero and the emphasis image e (x, y) is flat. This is because the pixel value in the unit changes, and the color of the output image changes.

<About the effect of this embodiment>
According to this embodiment, in addition to the function of removing the periodic pattern causing the moiré occurrence, the function of emphasizing the input image can be realized. In addition, since the emphasizing process is performed based on the band pass image b (x, y) from which the frequency component causing the moire generation is removed, the frequency component causing the moire generation is amplified by the emphasizing process. Does not occur. That is, according to the configuration of the present embodiment, the visibility of characters and edges is improved by emphasizing only the frequency components that do not cause moiré occurrence while removing only the frequency components that cause moiré occurrence. It can be done. Further, by using a filter coefficient of which the amplitude characteristic at direct current (0 cycle / mm) is substantially zero, such as the reference numeral 502 in FIG. 5, the overall tint of the image can be maintained.

Example 4
In the first embodiment, since the correlation determination is performed for each color component, the correlation determination result is different for each color component, and as a result, the degree of smoothing is also different for each color component. However, if the degree of smoothing differs depending on the color component, color misregistration may be noticeable in the edge portion or thin line portion.

  Therefore, in this embodiment, a band pass image is generated based on the luminance component (brightness component) of the color image, and the correlation determination is performed using the band pass image. Then, the obtained correlation determination result is commonly used for all color components, and smoothing is performed for each color component. As a result, the degree of smoothing becomes the same regardless of the color component, and the color shift in the edge portion and the thin line portion becomes inconspicuous. Further, since it is not necessary to have a filter processing unit and a correlation determination unit for each color component, the circuit scale can be made smaller and the mounting cost can be reduced compared to the configuration of the first embodiment.

<About periodic pattern removal processing>
Hereinafter, the periodic pattern removal process performed by the periodic pattern removal circuit 230 in the present embodiment will be described using FIG. FIG. 8 is a block diagram showing an example of the periodic pattern removal circuit 230 in the present embodiment.

  The periodic pattern removal circuit 230 shown in FIG. 8 is different from the above-described embodiment (see FIGS. 3, 6, and 7), and has a configuration for performing processing for all color components. The periodic pattern removal circuit 230 includes a filter processing unit 802, a correlation determination unit 803, a smoothing unit 804, a smoothing unit 805, and a smoothing unit 806. These are similar to the filter processing unit 301, the correlation determination unit 302, and the smoothing unit 303 in the first embodiment. However, the periodic pattern removal circuit 230 in the present embodiment is different from the first embodiment in that a luminance component calculation unit 801 is provided.

  The luminance component calculation unit 801 generates a luminance image L (x, y) based on the pixel values of each color of the RGB input image, and outputs the luminance image L (x, y) to the filter processing unit 802. Assuming that the pixel values of each color of the RGB input image are i_R (x, y), i_G (x, y), i_B (x, y), for example, the pixel values of the luminance image L (x, y) Calculate according to 3). The method of deriving the luminance image is not limited to the method using the equation (3).

  The filter processing unit 802 generates a band pass image b (x, y) from the luminance image L (x, y), and the correlation determination unit 803 uses the band pass image to generate a correlation determination result c (x, y). To derive. The smoothing unit 804 performs a smoothing process on the input image i_R (x, y), the smoothing unit 805 performs a smoothing process on the input image i_G (x, y), and the smoothing unit 806 performs an input image A smoothing process is performed on i_B (x, y). At this time, the smoothing units 804, 805, and 806 perform the smoothing process on the input image based on the common correlation determination result c (x, y).

<About the effects of this embodiment>
According to this embodiment, it is possible to remove the periodic pattern that causes the occurrence of the moiré without generating the color shift in the edge portion or the thin line portion. In addition, the circuit scale can be reduced and the mounting cost can be reduced as compared with the configuration of the first embodiment.

  In the present embodiment, although the correlation determination is performed using the luminance image, it is preferable that the correlation determination result does not differ for each color plane, so after performing the correlation determination for each color plane, based on the correlation determination result for each color plane Thus, the correlation determination result used for the smoothing process may be derived. Thus, the effect of suppressing color misregistration can be enhanced as compared to the case of using only a luminance image.

For example, the correlation determination result used for the smoothing process may be determined by majority decision. Here, the correlation determination results of R, G, and B are c_R (x, y), c_G (x, y), c_B (x, y), and these pixel values in the region of interest 403 are c_R _ij, c_G _ij , C_B _ij . Further, the correlation determination result used for smoothing is c (x, y), and the pixel value of c (x, y) in the focused area 403 is c _ij . In the case where c _ij is determined by majority, if any two or more of c_R _ij, c_G _ij , and c_B _ij at the position of (i, j) in the region of interest 403 become 1 (correlated), The pixel value c _ij at the same position (i, j) is 1 (correlated). On the other hand, when any two or more of c_R _ij, c_G _ij and c_B _ij become 0 (no correlation), the pixel value c _{ij is set} to 0 (no correlation).

Alternatively, even in the smoothing priority mode, the logical sum of the correlation determination result for each color plane is used, and in the sharpening priority mode, the logical product of the correlation determination result for each color plane is used, depending on the mode. good. That is, in the smoothing priority mode, when one or more of c_R _ij, c_G _ij , and c_B _ij at the position of (i, j) in the target area 403 become 1 (correlated), the same position (i the pixel value c _ij in j) 1 (Yes correlation) to. On the other hand, in other cases, it is 0 (no correlation). In the sharpening priority mode, when c_R _ij, c_G _ij and c_B _ij all become 1 (correlated), the pixel value c _ij at the same position (i, j) is 1 (correlated), In other cases, it is 0 (no correlation).

Alternatively, by adding the correlation determination result for each color plane, the value of each pixel can be multi-valued (for example, 0, 1, 2, 3 in the case of R, G, B) according to the number of color planes Common correlation determination results may be derived. Specifically, the sum of c_R _ij, c_G _ij and c_B _{ij at} the position (i, j) of the region of interest 403 is taken as the pixel value c _ij at the same position (i, j). Smoothing is performed based on _ij . In this case, the smoothing unit for each color component performs the smoothing to an extent according to the pixel value c _ij of the common correlation determination result. For example, as in a known bilateral filter, the sum of the pixel value i _ij in the target area 401 of the input image i (x, y) and the value obtained by multiplying the pixel value c _ij for each pixel is the pixel value c _ij Divide by the sum. Then, the value after this division is used as the pixel value after smoothing of the target pixel.

  In addition, the forms of Embodiments 1 to 4 described above may be used in combination as appropriate.

[Other embodiments]
The present invention supplies a program that implements one or more functions of the above-described embodiments to a system or apparatus via a network or storage medium, and one or more processors in a computer of the system or apparatus read and execute the program. Can also be realized. It can also be implemented by a circuit (eg, an ASIC) that implements one or more functions.

Claims (15)

A filter processing unit that generates a band pass image by performing filter processing that extracts a component of a predetermined spatial frequency range based on an input image;
A determination unit that determines whether there is a correlation between the pixel of interest and the peripheral pixels of the pixel of interest based on the band pass image;
Smoothing means for performing smoothing based on the input image according to the correlation determination result output by the determination means;
And a halftone processing unit configured to generate a dot pattern based on an output image of the smoothing unit.   The image processing according to claim 1, wherein the predetermined spatial frequency range is a range excluding a frequency component that causes moire due to interference with the dot pattern and a low frequency component near a direct current. apparatus.   The image processing apparatus according to claim 1, wherein the maximum value of the amplitude characteristic of the filter processing unit is 1 or less. The halftone processing means performs halftone screen processing with a predetermined number of lines,
The image processing apparatus according to any one of claims 1 to 3, wherein the upper limit of the predetermined spatial frequency range is set higher as the predetermined number of lines is higher.   According to another aspect of the present invention, the image processing apparatus further comprises setting means for deriving change amounts of pixel values in a target area of the band pass image, and setting a threshold for correlation determination used in the determination means according to the derived change amounts. An image processing apparatus according to any one of Items 1 to 4.   The image processing apparatus according to claim 5, wherein the setting unit sets a threshold for the correlation determination so as to suppress the smoothing in the smoothing unit as the amount of change increases. Deriving means for deriving an emphasis amount for each pixel by multiplying the pixel value of the band pass image by a gain coefficient;
The image processing apparatus further comprises generation means for generating an emphasized image in which the sharpness is emphasized by adding the emphasis amount to pixel values of the input image,
The image processing apparatus according to any one of claims 1 to 4, wherein the smoothing unit performs the smoothing on the enhanced image. The input image includes a plurality of color components,
5. The smoothing method according to claim 1, wherein the smoothing unit for each color component performs the smoothing for each color component in accordance with a correlation determination result commonly used for all of the plurality of color components. An image processing apparatus according to any one of the preceding claims. The image processing apparatus further comprises generation means for generating a luminance image by calculating the luminance for each pixel based on the pixel value of each of the plurality of color components.
The filter processing unit generates the band pass image by performing the filter processing on the luminance image.
The present invention is characterized in that the determination means outputs the correlation determination result derived based on the generated band pass image to the smoothing means for each color component as the commonly used correlation determination result. 8. The image processing apparatus according to 8. The determination unit for each color component calculates a difference absolute value between the pixel of interest and the peripheral pixels, and when the calculated difference absolute value is less than or equal to a predetermined threshold, the pixel of interest and the peripheral pixels It is determined that the correlation is between
The image processing apparatus according to claim 8, wherein the correlation determination result commonly used is derived based on the correlation determination result for each color component.   The image processing apparatus according to claim 10, wherein the commonly used correlation determination result is derived by a majority vote using the correlation determination result for each color component.   The image processing apparatus according to claim 10, wherein the commonly used correlation determination result is a logical sum or a logical product of the correlation determination result for each of the color components. The correlation determination result commonly used is derived by adding the correlation determination result for each color component,
The image processing apparatus according to claim 10, wherein the smoothing unit for each color component performs the smoothing to an extent according to a pixel value of the commonly used correlation determination result. A generation step of generating a band pass image by performing a filter process of extracting a component of a predetermined spatial frequency range based on an input image;
A determination step of determining whether there is a correlation between the pixel of interest and the peripheral pixels of the pixel of interest based on the band pass image;
A smoothing step for performing smoothing based on the input image according to the correlation determination result in the determination step;
And D. generating a dot pattern based on the output image in the smoothing step.   The program for making a computer perform the method of Claim 14.