JP3966440B2 - Image processing method, image processing apparatus, and storage medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to various apparatuses that handle multi-gradation image data such as digital copying machines, printers, fax machines, and displays, and more particularly to an image processing method and apparatus that uses an error diffusion method for quantization of multi-gradation image data. .
[0002]
[Prior art]
As a typical halftone processing method in an image processing apparatus related to image formation, there are a dither method and an error diffusion method.
[0003]
The dither method has an advantage that it has excellent graininess and can express a halftone image smoothly, but it also has a disadvantage. For example, resolution is deteriorated to obtain gradation. In addition, in the dither method for generating a periodic image, moire tends to occur on a printed image such as a halftone dot.
[0004]
On the other hand, the error diffusion method can obtain a resolution that is faithful to the original image and is suitable for reproducing a character image. However, in a halftone image such as a photograph, isolated dots are dispersed or irregularly connected to each other, so that granularity is poor and a peculiar texture may occur. In addition, in an electrophotographic printer, the image is unstable because an image is formed with isolated dots, and in error diffusion, the ratio of the small dots increases, so the stability further decreases, and graininess due to density unevenness Deterioration and banding are likely to occur.
[0005]
Regarding the error diffusion method, in order to improve the texture due to the irregular connection of dots, the dither threshold is used as the quantization threshold, and the following improvements including the method of improving the texture by disturbing the dot connection Technology has been proposed.
(1) A dither threshold is used for the purpose of eliminating the occurrence of pseudo contours and unique stripe patterns, and the amount of error diffusion is increased as the edge amount increases (Japanese Patent Laid-Open No. 3-34772).
(2) For the purpose of preventing white spots in a non-edge low density portion and preventing occurrence of a character notch, a fixed threshold is used in the edge portion of the image, a variation threshold is used in the non-edge portion, and the level of the variation threshold The lower the concentration, the lower the value (Japanese Patent No. 2755307).
(3) When using a multi-value printer of three or more values, a dither signal having a magnitude corresponding to the edge amount is added to the image data at the edge portion of the image for the purpose of preventing the occurrence of moire and pseudo contour. In the edge portion, a fixed value is added to the image data, and the added image data is subjected to multi-value quantization using a fixed threshold value (Japanese Patent No. 2801195).
[0006]
[Problems to be solved by the invention]
An object of the present invention is to form a high-quality image having excellent stability in a photograph, a flat image portion, a high-dot halftone dot image portion, etc. A point image section or the like is to provide an image processing method and apparatus for forming an image with high resolution.
[0007]
[Means for Solving the Problems]
In one aspect of the present invention, by performing edge detection of multi-tone image data and performing area expansion on the detection result, an edge portion of a character or line drawing, a halftone dot image portion having a relatively low number of lines, etc. Detect as an edge region. Also, a white background pixel and a non-white background pixel are discriminated. The multi-tone image data is quantized by the error diffusion method. As the quantization threshold, a fixed value is basically used for pixels in the edge region, and oscillation is periodically performed for pixels in the non-edge region. A quantization threshold is used. By doing so, a photograph, an image flat portion, a high-line number halftone dot image portion, etc. can form an image with excellent dither tone graininess and stability, and an edge portion of a character or line drawing, relatively A low-line-number halftone dot image portion and the like can form an image with excellent resolution of error diffusion tone. However, as will be described later in connection with the embodiment, a white background flat portion is associated with region expansion. Bordered noise may occur. In order to prevent the occurrence of such noise, a fixed value is used as a quantization threshold for pixels that can be regarded as a white background portion even in a non-edge region.
[0008]
According to another aspect, Similarly, the quantization threshold is fixed for the pixels in the edge region and the quantization threshold is periodically oscillated for the pixels in the non-edge region, but the quantization data is forced to a white value for the pixels in the white background portion. By setting, the generation of the above-described unnecessary noise is prevented. Depending on the configuration of the means for forcibly setting the white value, it is possible that the error accumulated by the white value setting is delayed and the density generation at the low density image start portion may be delayed. In this case, in order to prevent such a delay in density generation, the quantization error calculated for the white background pixel is replaced with a random value.
[0009]
According to another aspect of the present invention, in order to prevent generation of unnecessary noise in the flat portion of the white background related to the region expansion, the pixel of the white background portion is not based on the result of the region expansion but on the result of the edge detection. To control the quantization threshold.
[0010]
In another aspect of the present invention, in order to form a smooth image in which the boundary between the edge portion and the non-edge portion is aligned without a sense of incongruity, the edge degree of each pixel is detected instead of the edge / non-edge in the edge detection. The vibration width of the quantization threshold is controlled according to the edge degree, and details thereof will be described later in connection with the embodiment.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the accompanying drawings. In addition, in order to avoid duplication of description, the same reference number is used for the same part or a corresponding part in several drawings in the attached drawings.
[0012]
Example 1
FIG. 1 shows a block configuration of an image processing apparatus according to Embodiment 1 of the present invention. This image processing apparatus receives multi-gradation image data 100 and outputs quantized data 101. The image processing apparatus includes a quantization processing unit 120, an image feature extraction unit 130, a quantization threshold generation unit 140, and a quantization processing unit. 120 includes a signal delay unit 150 for adjusting timing between the image feature extraction unit 130 and the image feature extraction unit 130, a white background determination unit 160, and a selection unit 161. The signal delay unit 150 is provided as necessary, and is a line memory having a required number of lines, for example. The input image data 100 is 8-bit / pixel data read by a scanner at 600 dpi, for example. In general, such image data 100 is input after passing through a smoothing filter in order to smoothly express a halftone. Usually, since smoothing is performed from an image period of about 150 Lpi, the periodic component of a high-line number halftone dot image of 175 Lpi or more used in gravure printing or the like does not remain in the image data 100.
[0013]
The quantization processing unit 120 quantizes the multi-tone image data by the error diffusion method using the quantization threshold given from the selection unit 161. In this embodiment, as shown in the figure, the quantizer (Comparator) 121, error calculation unit 122, error storage unit 123, error diffusion matrix unit 124, and error addition unit 125. The image data 100 is adjusted in timing by the signal delay unit 150 and input to the error addition unit 125. The image data added with the diffusion error by the error adder 115 is input to the quantizer 121. The quantizer 121 quantizes the input image data using a quantization threshold, and outputs the quantization result as quantized data 101.
[0014]
In order to simplify the description, in this embodiment and each embodiment described later, only one quantization threshold is supplied via the selection unit 161, and the quantizer 121 receives the input image data as a quantization threshold. In the above description, it is assumed that 1-bit quantized data 101 having a value of “1” and otherwise “0” is output. However, the present invention is not limited to this. For example, a configuration may be adopted in which three quantization thresholds are generated, and the quantizer 121 uses the quantization thresholds to quantize the image data into four levels and output 2-bit quantized data 101. it can.
[0015]
The error calculation unit 122 calculates the quantization error of the quantizer 121. Since 8-bit image data is handled here, in this error calculation, for example, “1” of the quantized data 101 is treated as 255 (decimal) and “0” is treated as 0 (decimal). The calculated quantization error is temporarily stored in the error storage unit 123. The error storage unit 123 is for storing a quantization error related to a processed pixel around the target pixel. In this embodiment, as will be described below, for example, a two-line line memory is used as the error storage unit 123 in order to diffuse the quantization error to surrounding pixels two lines ahead.
[0016]
The error diffusion matrix unit 124 calculates a diffusion error to be added to the next pixel of interest from the quantization error data stored in the error storage unit 123. In this embodiment, the error diffusion matrix unit 125 calculates diffusion error data using an error diffusion matrix having a size of 3 pixels in the sub-scanning direction and 5 pixels in the main scanning direction as shown in FIG. In FIG. 2, the mark * corresponds to the position of the next pixel of interest, and a, b,. . . , K, l are coefficients corresponding to the positions of the 12 processed pixels in the vicinity (the sum is 32). In the error diffusion matrix unit 125, a value obtained by dividing the product sum of the quantization error for the 12 processed pixels and the corresponding coefficients a to l by 32 is sent to the error addition unit 125 as a diffusion error for the next pixel of interest. give.
[0017]
The image feature extraction unit 130 includes an edge detection unit 131 and a region expansion processing unit 132. The edge detection unit 131 performs edge detection of the image data 100. In another aspect of the present invention, the edge detection unit 131 discriminates only the edge / non-edge, but in the present embodiment and each embodiment described later, the edge detection unit 131 detects the edge degree of each pixel. In the present embodiment, the edge detection unit 131 quantizes the edge degree into nine levels from level 0 (maximum edge degree) to level 8 (minimum edge degree, non-edge), and outputs it as 4-bit edge data. To do. More specifically, for example, the four types of 5 × 5 differential filters shown in FIG. 3 are used to detect the edge amount in the four directions of the main scanning direction, the sub-scanning direction, and the direction inclined by ± 45 ° from the main scanning direction. Then, the edge amount having the maximum absolute value is selected, and the absolute value of the edge amount is quantized to 9 level edge levels from level 0 to level 8 and output.
[0018]
The region expansion processing unit 132 performs region expansion of the edge detected by the edge detection unit 131, and in this embodiment, the region expansion width is 7 pixels wide. That is, with reference to the edge data output from the edge detection unit 131, the minimum in the 7 × 7 pixel area (range of 3 pixels before and after the main scanning direction and 3 pixels before and after the sub scanning direction) around the target pixel The edge level (maximum edge degree) is set as the edge level of the pixel of interest, and this is output as 4-bit edge data. This edge data is given to the quantization threshold value generator 140.
[0019]
The quantization threshold generation unit 140 generates a quantization threshold that periodically vibrates on the image space with a vibration width corresponding to the edge level represented by the edge data output from the region expansion processing unit 132. A dither threshold value generator 141, a multiplier 142 that multiplies the output value of the dither threshold value generator 141 by a coefficient (0 to 8) corresponding to the edge level indicated by the edge data, and a fixed value for the output value of the multiplier 142. It comprises an adding unit 143 for adding.
[0020]
In this embodiment, the dither threshold value generator 141 spirals clockwise from the inside to the outside in order from the smallest threshold values from −7 to +8 as shown in FIG. 4 (−7 is minimum and +8 is maximum). Using the arranged 4 × 4 dither threshold matrix, a dither threshold that oscillates periodically from −7 to +8 in the image space is output. The dither threshold period is 4 pixels, which corresponds to 150 Lpi in the case of 600 dpi image formation. Such a dither threshold value generation unit 141 can be easily realized by a ROM storing the dither threshold value matrix, a counter that counts main and sub-scan timing signals of image data, and generates a read address of the ROM. .
[0021]
The multiplication unit 142 sets the coefficient 8 when the edge level indicated by the edge data from the image feature extraction unit 130 is level 8 (non-edge), the coefficient 7 when it is level 7, the coefficient 6 when it is level 6, and the level 5 Factor 5 at level 4, factor 4 at level 4, factor 3 at level 3, factor 2 at level 1, factor 1 at level 1, factor 0 at level 0 (edge degree maximum) The output value of the dither threshold value generator 141 is multiplied. Therefore, the output value of the multiplication unit 142 vibrates with the maximum vibration width from +64 to −56 at the edge level 8 (non-edge). The fixed value added by the adding unit 143 is selected as the median value of the image data width +128 (decimal). Therefore, the quantization threshold given to the quantizer 121 vibrates around +128, and the maximum vibration width is 120 (from +192 to 72).
[0022]
The white background determination unit 160 determines a white background pixel and a non-white background pixel of the image data 100. As a determination method, a method of determining a white background pixel when the image data level of the target pixel is equal to or less than a predetermined determination threshold value, and a non-white background pixel when exceeding the determination threshold value can be used. However, in order to enable correct determination even when the image data level locally exceeds the determination threshold within the white background portion, for example, the average value of the image data level of the 3 × 3 region centered on the target pixel is determined. It is preferable to use a determination method that compares the threshold value with a white background pixel when the average value is equal to or less than the determination threshold value, and sets the non-white background pixel when the average value exceeds the determination threshold value. In the case where the image data 100 has 256 gradations having values from 0 to 255, for example, 5 or a slightly larger or slightly smaller value can be used as the determination threshold.
[0023]
The determination output of the white background determination unit 160 is given to the selection unit 161. When this determination output indicates a pixel in a non-white background portion, the selection unit 161 selects the quantization threshold generated by the quantization threshold generation unit 140 and uses the quantization threshold as a quantization threshold. However, when the determination output indicates a white background pixel, the selection unit 161 selects a predetermined fixed value (in this case, +128) and supplies it to the quantizer 121 as a quantization threshold. Such quantization threshold switching control is performed in order to suppress generation of unnecessary noise related to area expansion as described later.
[0024]
If the quantized data 101 of the image processing apparatus configured as described above is given to, for example, an electrophotographic printer, the character, the changing point of the image, the halftone dot image portion having a relatively low number of lines, and the like can be resolved. In addition, a photograph, a portion with little change in the image, a halftone dot image with a high number of lines, etc. are smooth and stable, and it is possible to form a high-quality image in which these regions are aligned without a sense of incompatibility. This will be described below.
[0025]
Here, in order not to make the description complicated, it is assumed that the quantization threshold generated by the quantization threshold generation unit 140 is selected by the selection unit 161.
[0026]
In an edge region where the change is steep and the edge level is level 0 (the edge degree is the highest) such as the edge of a character or line drawing in the image, the quantization threshold generated by the quantization threshold generator 140 is fixed to +128. Therefore, since the quantization processing unit 120 performs quantization processing by a pure error diffusion method using a fixed threshold value, it is possible to form an error diffusion tone image with high resolution.
[0027]
In a non-edge region having a minimum edge degree (maximum edge level) such as a flat part of a photograph or image, the oscillation width of the quantization threshold generated by the quantization threshold generation unit 140 is maximized. The quantization process of the unit 120 is a dither tone process, and the image data is halftoned at a dither threshold period to form an image with good graininess and stability.
[0028]
In the image processing apparatus according to the present embodiment, the vibration width of the quantization threshold is gradually increased or decreased at the boundary between the edge region and the non-edge region according to the degree of the edge. The characteristics of the quantization process can be smoothly switched in the opposite direction or vice versa. Therefore, it is possible to form an image in which the boundary portion between the edge region and the non-edge region is aligned without a sense of incongruity.
[0029]
Now, the region expansion processing unit 132 of the image feature extraction unit 130 performs a region expansion of 7 pixels on the edge detection result. In the case of 600 dpi image formation, the region expansion width of 7 pixels is approximately about the image space. At 0.3 mm, this corresponds to a halftone dot period of about 86 Lpi. Accordingly, a relatively low line number halftone dot image portion of 86 Lpi or more is evaluated as an edge region, and error diffusion processing using a fixed quantization threshold is performed in the quantization processing unit 120. It can be faithfully reproduced at high resolution and no moire occurs.
[0030]
In addition, as described above, a dot component having a high line number of 175 Lpi or more is smoothed and does not remain in the image data 100. Therefore, such a high dot number halftone dot image portion has an edge level of level 8 or a high level. As in the case of the flat image portion, the image is re-dotted with the dither threshold period (150 Lpi) by the dither tone processing using the quantization threshold having a large vibration width, and an image having excellent graininess and stability can be formed. Further, since the halftone dot component is lost from the image data 100, moire does not occur.
[0031]
In a halftone dot image with a low line number of less than 86 Lpi, the halftone dot boundary portion evaluated as an edge is processed mainly by error diffusion using a fixed quantization threshold, so that the halftone dot is reproduced faithfully and moiré is generated. In addition, the halftone dot center portion that is not evaluated as an edge region is subjected to dither tone processing using the oscillating quantization threshold, and thus can form an image with good stability and granularity.
[0032]
Next, the operation of the selection unit 161 will be described with reference to FIG. FIG. 5A shows an example of a typical original image in which unnecessary noise is likely to occur due to the extension of the edge region in the image feature extraction unit 130. This original image is an image in which a relatively low number of dot portions, edge portions, and white background portions, which are evaluated as edge regions by region expansion, are arranged in this order. When the quantization threshold value switching control by the selection unit 161 is not performed, when the multi-tone image data of the original image in FIG. 5A is processed, the white background portion of the output image is trimmed as shown in FIG. -Like unwanted dots appear. The cause is as follows.
[0033]
By the region expansion processing of the region expansion processing unit 132, the range of 7 lines near the edge portion of the white background portion of the original image is evaluated as an edge region, and error diffusion processing using a fixed quantization threshold is performed. From the next line, the process switches to a dither keying process using a quantization threshold that vibrates with a large vibration width. In a situation where a large amount of positive quantization error is accumulated by error diffusion processing in the original edge region (the edge portion shown in FIG. 5A), the quantizer is added by adding the error even if it enters the white background portion. The input image data 121 becomes a relatively high level. In the 7 lines closer to the edge portion that are defined as edge regions by region expansion, the quantization threshold value is fixed to +128, so that the image data level does not exceed the quantization threshold value. However, since the quantization threshold vibrates with a large vibration width at the 8th line that switches to the dither tone processing, when the quantization threshold becomes a small value, the image data level after error addition exceeds the quantization threshold and is quantized. The converted data 101 becomes “1”, which is output as an unnecessary dot. Such unnecessary dots are more likely to occur as the vibration width of the quantization threshold is larger. Also, as the number of quantization levels increases, unnecessary dots are more likely to occur because a lower level quantization threshold is used.
[0034]
In the image processing apparatus according to the present exemplary embodiment, when processing the white background portion, the selection unit 161 selects a fixed value +128 according to the determination output of the white background determination unit 160 and supplies the selected value to the quantizer 121 as a quantization threshold. To do. Accordingly, the image data level after error addition does not exceed the quantization threshold in the white background portion, and such unnecessary dots do not occur. In the original image shown in FIG. 5A, the upper portion of the edge portion is a halftone dot portion having a relatively low line number, but the same is true even if the portion is a flat portion or a high line number halftone dot portion. is there.
[0035]
According to the modification of the present embodiment, the dither threshold value generation unit 141 of the quantization threshold value generation unit 140 uses an 8 × 8 dither threshold value matrix as shown in FIG. This dither threshold matrix is a combination of four 4 × 4 basic dither threshold matrices (same as the dither threshold matrix shown in FIG. 4) surrounded by a thick line frame and expanded to 8 × 8. The basic dither threshold value matrix adjacent in the scanning direction is relatively half-phase shifted in the sub-scanning direction. If such a dither threshold matrix is used, a screen angle of approximately 63.5 ° is added to the halftone dot arrangement in an image flat portion or the like which is a dither tone process. There is little collapse and the graininess is good.
[0036]
The dither threshold value generator 141 does not necessarily have a ROM that stores an 8 × 8 dither threshold matrix shown in FIG. 6, but has a ROM that stores a 4 × 4 basic dither threshold matrix, and has a read address of the ROM. By the control, it is possible to generate a quantization threshold by the dither threshold matrix of FIG.
[0037]
Example 2
According to the second embodiment of the present invention, the quantization threshold value generation unit 140 is configured as shown in FIG. 7 in the image processing apparatus having the overall configuration shown in FIG. The edge detection unit 131 of the image feature extraction unit 130 (FIG. 1) quantizes the edge degree into four edge levels from level 0 (maximum edge degree) to level 3 (non-edge), and converts it into a 2-bit edge. Changed to output as data.
[0038]
As shown in FIG. 7, the quantization threshold value generation unit 140 in this embodiment includes a threshold value generation unit 145_0 corresponding to the edge level 0, a threshold value generation unit 145_1 corresponding to the edge level 1, and a threshold value generation corresponding to the edge level 2. 145_2, a threshold generation unit 145_3 corresponding to edge level 3, and a threshold generated by any one of threshold generation units 145_0 to 145_3 according to the edge level indicated by the edge data output from the image feature extraction unit 130 , And a threshold selection unit 146 that gives the quantization threshold to the quantizer 121 of the quantization processing unit 120 (FIG. 1).
[0039]
The threshold value generator 145_3 corresponding to the edge level 3 (non-edge) uses, for example, a dither threshold value matrix obtained by multiplying each threshold value of the dither threshold value matrix shown in FIG. 4 (or FIG. 6) by 8 and then adding 128. A threshold value that vibrates at the maximum vibration width is generated. The threshold generation unit 145_2 corresponding to the edge level 2 uses a dither threshold matrix obtained by multiplying each threshold of the dither threshold matrix shown in FIG. 4 (or FIG. 6) by 5 and adding 128, for example. A threshold value that vibrates is generated. The threshold generation unit 145_1 corresponding to the edge level 1 uses a dither threshold matrix obtained by multiplying each threshold of the dither threshold matrix shown in FIG. 4 (or FIG. 6) by 2 and adding 128, for example. A threshold value that vibrates is generated. The threshold value generator 145_0 corresponding to the edge level 0 (edge degree maximum) generates a fixed value (+128).
[0040]
Therefore, it is obvious that a high-quality image can be formed in the present embodiment by performing the same quantization process as in the first embodiment (or its modification).
[0041]
According to the configuration of the quantization threshold value generation unit 140 of the present embodiment, a means for multiplication that is generally disadvantageous in terms of cost or processing time (multiplication in FIG. 1) is realized in either hardware or software. Can be eliminated. The area expansion processing unit 132 needs to temporarily store edge data for a plurality of lines corresponding to the area expansion width. However, the edge data is temporarily stored by the amount corresponding to the compression of the edge data to 2 bits. The capacity of the line memory can be reduced. In addition, since the number of edge levels is as small as four, the threshold generation units 145_0 to 145_2 need a small amount of memory for storing the dither threshold matrix.
[0042]
Although the memory of the area expansion processing unit 132 is not reduced, a configuration in which the edge detection unit 131 outputs 9 levels of the edge level and the area expansion unit 132 converts the edge level to 4 levels and outputs it is also possible.
[0043]
Example 3
FIG. 8 is a block diagram of an image processing apparatus according to the third embodiment of the present invention. The same reference numerals in FIG. 8 as those in FIG. 1 denote the same parts. The quantization threshold value generator 140 may have the configuration shown in FIG. 1 or the configuration shown in FIG.
[0044]
In this embodiment, an output value control unit 165 is added after the quantization processing unit 120. The output value control unit 165 outputs the quantized data 101 output from the quantization processing unit 120 as it is as the quantized data 102 when the determination output of the white background determination unit 160 indicates a non-white background pixel. To do. However, when the determination output of the white background determination unit 160 indicates a pixel in the white background portion, the output value control unit 165 outputs the quantized data 102 set to the white value regardless of the value of the quantized data 101. The quantized data 102 is supplied to an electrophotographic printer or the like for image formation.
[0045]
Because of this configuration, for example, when multi-tone image data of the original image as shown in FIG. 5A is processed, the quantized data 102 is forcibly set to a white value (0) in the white background portion. Therefore, an unnecessary dot in the form of a border as shown in FIG. 5B does not occur in the portion where the dither tone processing is switched in the white background portion.
[0046]
Example 4
FIG. 9 shows a block configuration of an image processing apparatus according to Embodiment 4 of the present invention. The quantization threshold value generator 140 may have the configuration shown in FIG. 1 or the configuration shown in FIG.
[0047]
In this embodiment, the determination output of the white background determination unit 160 is given to the quantizer 121 of the quantization processing unit 120. The quantizer 121 outputs the white value (0) as the quantized data 101 regardless of the image data level for the pixels determined as the white background pixels by the white background determination unit 160.
[0048]
In addition, the quantization processing unit 120 includes a random number generation unit 127 that generates a random value and an error written in the error storage unit 123 by the quantization error calculated by the error calculation unit 122 or the random number generation unit 127. A selection unit 126 for selecting a random value is added. When the determination output of the white background determination unit 160 indicates a non-white background pixel, the selection unit 126 selects the quantization error calculated by the error calculation unit 122. However, when the determination output of the white background determination unit 160 indicates a white background pixel, the random value generated by the random number generation unit 127 is selected by the selection unit 126. That is, when the target pixel is a white background pixel, the quantization error written in the error storage unit 123 is replaced with a random value.
[0049]
Because of this configuration, for example, when multi-tone image data of the original image as shown in FIG. 5A is processed, the quantized data 101 is forcibly set to a white value (0) in the white background portion. Therefore, an unnecessary dot in the form of a border as shown in FIG. 5B does not occur in a portion where the dither tone processing is switched in the white background portion.
[0050]
For the white background pixels, a random value is selected by the selection unit 126 and written to the error storage unit 123. If such replacement with random values is not performed, the forced white value setting of the quantized data 101 does not proceed with the elimination of accumulated errors, and a phenomenon occurs in which the density generation at the low density image start portion is delayed. there is a possibility. By replacing the quantization error with a random value, such a phenomenon can be prevented.
[0051]
Example 5
FIG. 10 shows a block configuration of an image processing apparatus according to Embodiment 5 of the present invention. 10, the same reference numerals as those in FIG. 1 denote the same parts. However, the quantization threshold value generator 140 may have the configuration shown in FIG. 1 or the configuration shown in FIG.
[0052]
In the present embodiment, a selection unit 167 is added. When the determination output of the white background determination unit 160 indicates a non-white background pixel, the selection unit 167 sends the edge data output from the region expansion processing unit 132 of the image feature extraction unit 130 to the quantization threshold generation unit 140. give. Therefore, as in the above embodiments, the vibration width of the quantization threshold is controlled according to the edge degree after the region expansion process. However, when the determination output of the white background portion determination unit 160 indicates a pixel of the white background portion, the edge data output from the edge detection unit 131 of the image feature extraction unit 130 by the selection unit 167 is given to the quantization threshold value generation unit 140. Therefore, the vibration width of the quantization threshold is controlled according to the edge degree before region expansion.
[0053]
According to such a configuration, for example, when multi-tone image data of an original image as shown in FIG. 5A is processed, a quantization threshold value that vibrates immediately with a large vibration width when entering the white background portion from the edge portion. It switches to the dithering process using, and dots tend to occur. As a result, even if a lot of positive errors are accumulated in the processing of the edge portion, dots are output in the vicinity of the edge portion, and an unpleasant bordered unnecessary dot as shown in FIG. 5B is generated. No longer.
[0054]
As described above, in the same configuration as each of the above embodiments, the image feature extraction unit outputs two-level edge data indicating edge / non-edge, and the edge pixel has a fixed value and non-edge data. It can also be used as a strong vibration quantization threshold for a pixel. Even in this way, the edges of characters and line drawings, and halftone dot image portions with a relatively low number of lines have high resolution, and the photograph, flat image portion, and high line number halftone dot image portions have graininess and stability. Therefore, it is possible to form a good image, and it is possible to prevent the generation of unnecessary dots in the form of an edge in the white background as described with reference to FIG.
[0055]
In addition, it is possible to execute image processing equivalent to that of the image processing apparatus of each embodiment by a program using a general-purpose or dedicated computer. For example, a program for that purpose is read from various storage media such as a magnetic disk, an optical disk, a magneto-optical disk, and a semiconductor storage element on which the program is recorded, or is received from an external computer via a network, and the main computer Various storage media in which such a program is recorded are also included in the present invention. Next, an embodiment in which processing is executed by a program using a computer will be described.
[0056]
Example 6
FIG. 11 is a schematic flowchart of a program according to the sixth embodiment of the present invention. In this embodiment, image processing equivalent to that in the first or second embodiment is performed. In the following description, it is assumed that the entire multi-tone image data to be processed is stored in advance in a frame memory area on the main memory of the computer. It is also assumed that a storage area for storing quantized data, quantization error, edge data, etc. is also secured on the main memory.
[0057]
First, a target pixel is selected (step 200). Next, the same determination as that of the white background determination unit 160 is performed by comparing the image data level of the target pixel with a predetermined determination threshold (step 204).
[0058]
When it is determined that the pixel of interest is a white background pixel, a predetermined fixed value (for example, +128) is set as the quantization threshold (step 212), and the quantization threshold is used in the same manner as the quantization processing unit 120. The quantization process by the error diffusion method is performed on the target pixel (step 214). The quantized data and the calculated quantization error are accumulated in the respective storage areas.
[0059]
If it is determined in step 204 that the target pixel is a non-white background pixel, edge detection similar to the edge detection unit 131 is performed (step 206), and then the region expansion processing unit 132 is performed. Region expansion processing is performed (step 210). Since the edge data within the range of 7 × 7 is referenced for the area expansion process, the edge data for 7 lines obtained by the edge detection is temporarily stored in the storage area for that purpose. Then, according to the edge data of the target pixel obtained by the region expansion process, the quantization threshold is determined by the same algorithm as that of the quantization threshold generation unit 140 (step 210). However, as described above, multi-level edge degree detection is not performed, only edge / non-edge detection is performed, a quantization threshold is fixed for edge pixels, and a high-amplitude dither threshold is set for non-edge pixels. You may make it do. Next, the pixel of interest is quantized by the error diffusion method using this quantization threshold, and the obtained quantized data and quantization error are accumulated in the respective storage areas (step 214).
[0060]
When the processing of one target pixel is completed as described above, the next pixel is selected as the target pixel (step 200), and the processing after step 204 is executed. When the same processing is repeated and the processing is completed up to the last pixel of the frame (step 202, YES), the quantized data stored in the main memory is output to, for example, an external printer, a computer hard disk, or the like (step). 216).
[0061]
Example 7
FIG. 12 is a schematic flowchart of a program in the seventh embodiment of the present invention. In the present embodiment, image processing equivalent to that in the fourth embodiment is performed. In the following description, it is assumed that the entire multi-tone image data to be processed is stored in advance in a frame memory area on the main memory of the computer. It is also assumed that a storage area for storing quantized data, quantization error, edge data, etc. is also secured on the main memory.
[0062]
First, a target pixel is selected (step 250). Next, the same determination as that of the white background determination unit 160 is performed by comparing the image data level of the target pixel with a predetermined determination threshold (step 254).
[0063]
If it is determined in the white background determination (step 254) that the pixel of interest is a non-white background pixel, edge detection similar to the edge detection unit 131 is performed (step 256), and then the area expansion processing unit 132 and Similar area expansion processing is performed (step 258). According to the edge data of the target pixel obtained by this region expansion processing, the quantization threshold is determined by the same algorithm as that of the quantization threshold generation unit 140 (step 260). Then, quantization processing by the error diffusion method is performed using the quantization threshold value, and the quantized data and the quantization error are accumulated in the respective storage areas (step 262). It should be noted that, as described above, it is possible to perform edge detection of only edges / non-edges in step 256, and to set the quantization threshold to a fixed value for edge pixels and to a strong amplitude dither threshold for non-edge pixels. is there.
[0064]
If it is determined in step 254 that the pixel of interest is a white background pixel, a white value (0) is written in the quantized data storage area as the quantized data of the pixel of interest (step 264). As a result, a random value is written in the error storage area (step 266).
[0065]
When the process related to the target pixel is completed in this way, the next pixel is selected as the target pixel (step 250), and the processes after step 254 are executed. Thereafter, the same process is repeated, and when the process is completed up to the final pixel of the frame (YES in step 252), the quantized data stored in the storage area on the main memory is output to, for example, an external printer or a computer hard disk (Step 268).
[0066]
Example 8
FIG. 13 is a schematic flowchart of a program according to the eighth embodiment of the present invention. In the present embodiment, image processing equivalent to that in the fifth embodiment is performed. In the following description, it is assumed that the entire multi-tone image data to be processed is stored in advance in a frame memory area on the main memory of the computer. It is also assumed that a storage area for storing quantized data, quantization error, edge data, etc. is also secured on the main memory.
[0067]
First, a target pixel is selected (step 300), and edge detection similar to that of the edge detection unit 131 is performed (step 304).
[0068]
Next, the same determination as that of the white background determination unit 160 is performed by comparing the image data level of the target pixel with a predetermined determination threshold (step 306). If it is determined that the pixel is a white background pixel, the quantization threshold is determined by the same algorithm as the quantization threshold generation unit 140 in accordance with the edge data obtained in step 304 (step 310). Then, quantization processing by the error diffusion method is performed using the quantization threshold value, and the quantized data and the quantization error are accumulated in the respective storage areas (step 312).
[0069]
If it is determined in step 306 that the pixel of interest is a non-white background pixel, a region expansion process similar to the region expansion processing unit 132 is performed (step 308). In accordance with the edge data of the target pixel obtained by this region expansion processing, a quantization threshold is determined by the same algorithm as that of the quantization threshold generation unit 140 (step 310), and quantization using the error diffusion method is performed using this quantization threshold. Processing is performed, and the quantized data and the quantization error are accumulated in the respective storage areas (step 312). As described above, it is possible to perform edge detection of only edges / non-edges in step 304 and set the quantization threshold to a fixed value for edge pixels and a strong amplitude dither threshold for non-edge pixels. is there.
[0070]
When the process related to the target pixel is completed in this way, the next pixel is selected as the target pixel (step 300), and the processes after step 304 are executed. Thereafter, the same processing is repeated, and when the processing is completed up to the final pixel of the frame (step 302, YES), the quantized data stored in the main memory is output to, for example, an external printer, a hard disk of a computer (step) 314).
[0071]
The image processing method or apparatus of the present invention can be installed in a device such as a printer or a display related to image formation, or a device such as a digital copying machine or a fax device related to both image reading and image formation. As an example of such an embodiment, an example of a digital copying machine to which the present invention is applied will be described below.
[0072]
Example 9
FIG. 14 is a schematic cross-sectional view showing a configuration example of an image reading mechanism and an image forming mechanism of a digital copying machine. This digital copying machine has a scanner unit 400 that optically scans and reads a document, a laser printer unit 411 as an image forming unit, and a circuit unit 550 (not shown) (FIG. 15).
[0073]
The scanner unit 400 illuminates a document placed on a flat document table 403 with an illumination lamp 502 and connects the reflected light image to an image sensor 507 such as a CCD via mirrors 503, 504 and 505 and a lens 506. At the same time, the image information of the original is read by sub-scanning the original by moving the illumination lamp 502 and the mirrors 503 to 505. The analog image signal output from the image sensor 507 is input to the circuit unit 550 (FIG. 15) and processed. Image data output from the circuit unit 550 is input to the laser printer unit 411.
[0074]
In the laser printer unit 411, the writing optical unit 508 converts the image data input from the circuit unit 550 into an optical signal, and exposes an image carrier made of a photoconductor, for example, a photoconductor drum 509, whereby an original image is obtained. An electrostatic latent image corresponding to is formed. For example, the writing optical unit 508 emits a laser beam whose intensity is modulated by driving a semiconductor laser by the light emission drive control unit with the image data, and deflecting and scanning the laser beam by the rotary polygon mirror 510, and an f / θ lens. Then, the photosensitive drum 509 is irradiated through the reflection mirror 511. The photosensitive drum 509 is driven to rotate by a drive unit and rotates clockwise as indicated by an arrow, and is uniformly charged by a charger 512 and then exposed by a writing optical unit 508 to form an electrostatic latent image. Is done. The electrostatic latent image on the photosensitive drum 509 is developed by the developing device 513 to become a toner image. Further, the paper is fed to the registration roller 520 from any of the plurality of paper feeding units 514 to 518 and the manual paper feeding unit 519. The registration roller 520 sends a sheet to the toner image on the photosensitive drum 509 in time. The transfer belt 521 is applied with a transfer bias from a transfer power source, transfers the toner image on the photosensitive drum 509 to a sheet, and conveys the sheet. The sheet onto which the toner image has been transferred is conveyed to the fixing unit 522 by the transfer belt 521 and the toner image is fixed, and then is discharged to the paper discharge tray 523. Further, the photosensitive drum 509 is cleaned by the cleaning device 524 after the toner image is transferred, and is further discharged by the charge eliminator 525 to prepare for the next image forming operation.
[0075]
FIG. 15 is a block diagram showing a simplified example of the circuit unit 550 of the digital copying machine. The input of the circuit unit 550 is an analog image signal read by the image sensor 507 of the scanner unit 400 at, for example, 600 dpi. This analog image signal is adjusted in level by the AGC circuit 551 and then converted into 8-bit digital image data per pixel by the A / D conversion circuit 552. Further, the shading correction circuit 553 further converts the analog image signal for each pixel of the image sensor 507. Variations in sensitivity and illuminance are corrected.
[0076]
The image data after the shading correction is sent to the filter processing circuit 556 and subjected to filter processing for MTF correction and smoothing. As described above, this filtering process smoothes halftone dot components with a line number higher than about 150 Lpi, and removes halftone dot components with a high line number higher than 175 Lpi almost completely. The filtered image data is sent to the gamma correction circuit 555 and subjected to gamma correction for conversion to writing density.
[0077]
Reference numeral 560 denotes a halftone processing unit. The halftone processing unit 560 includes the image processing apparatus of the present invention, for example, the image processing apparatus of the first, second, third, fourth or fifth embodiment. The image data after the filter processing is input to the image feature extraction unit (130) of the image processing device, and the image data after the gamma correction is input to the quantization processing unit (120) via the signal delay unit (150) of the image processing device. Entered. The quantized data output from the halftone processing unit 560 is sent to the light emission drive control unit of the semiconductor laser in the writing optical unit 508. However, the halftone processing unit 560 may be realized by a program using a computer as in the sixth, seventh, or eighth embodiment. In this case, the filter processing circuit 554 and the gamma correction color 555 can also be realized by a program.
[0078]
In the digital copying machine, the scaling process of the image data in the main scanning direction is performed, for example, before the gamma correction circuit 555, or the background removal processing or flare removal processing is performed, for example, with the gamma correction circuit 555. The processing may be performed in the middle of the processing unit 560, or the 90 ° rotation processing may be performed, for example, before the filter processing circuit 554 or after the halftone processing unit 560, but the description thereof is omitted.
[0079]
【The invention's effect】
As is clear from the above explanation, Claim 1, 3, 7 or 10 According to the invention, the change point of characters and images and the halftone dot image portion with a relatively low number of lines have high resolution, and the portion with little change in the photograph and image and the halftone dot image portion with a high number of lines have graininess. Therefore, it is possible to form an image having excellent stability, and it is possible to prevent the generation of an unnecessary dot in the form of a border having a sense of incongruity in the flat portion of the white background. Claim 2, 4, 8, 9, 11, or 12 According to the invention, the change point of characters and images and the halftone dot image portion with a relatively low number of lines have high resolution, and the portion with little change in the photograph and image and the halftone dot image portion with a high number of lines have graininess. In addition, it is possible to form a high-quality image that is excellent in stability and in which both areas are aligned with no sense of incongruity, and it is possible to prevent the generation of border-like unnecessary dots that have a sense of incongruity in the flat white portion. Claim 5 According to the described invention, even when there is noise in the white background portion, it is possible to reliably determine the pixels in the white background portion and perform appropriate processing. Claim 6 According to the described invention, it is possible to easily perform the image processing method of the present invention using a computer, and so on.
[Brief description of the drawings]
FIG. 1 is a block diagram of Embodiment 1 of the present invention.
FIG. 2 is a diagram illustrating an example of an error diffusion matrix.
FIG. 3 is a diagram illustrating an example of a differential filter for edge detection.
FIG. 4 is a diagram illustrating a dither threshold matrix used for quantization threshold generation in the first embodiment.
FIG. 5 is a diagram for explaining the generation of borderless unnecessary dots in a white background flat portion.
6 is a diagram showing a dither threshold value matrix used for generating a quantization threshold value in a modification of the first embodiment. FIG.
FIG. 7 is a block diagram of a quantization threshold value generation unit in Embodiment 2 of the present invention.
FIG. 8 is a block diagram of Embodiment 3 of the present invention.
FIG. 9 is a block diagram of Embodiment 4 of the present invention.
FIG. 10 is a block diagram of Embodiment 5 of the present invention.
FIG. 11 is a flowchart of Embodiment 6 of the present invention.
FIG. 12 is a flowchart of Embodiment 7 of the present invention.
FIG. 13 is a flowchart of Embodiment 8 of the present invention.
FIG. 14 is a schematic cross-sectional view showing a configuration example of a mechanism related to image reading and image formation of a digital copying machine according to Embodiment 9 of the present invention.
15 is a block diagram of a circuit unit of a digital copying machine according to Embodiment 9. FIG.
[Explanation of symbols]
100 Multi-tone image data
101 Quantized data
120 Quantization processor
121 Quantizer
122 Error calculator
123 Error storage unit
124 Error diffusion matrix part
125 Error adder
126 Selector
127 Random number generator
130 Image feature extraction unit
131 Edge detector
132 Area expansion processing unit
140 Quantization threshold value generator
141 Dither threshold generator
142 Multiplier
143 Adder
145_0 to 145_3 threshold generation unit
146 Threshold selection unit
150 Signal delay unit
160 White background part determination part
161 Selection part
165 Output value control unit
167 selection part
400 Scanner unit
411 Laser printer
551 AGC circuit
552 A / D conversion circuit
553 Shading correction circuit
556 Filter processing circuit
557 Gamma correction circuit
560 Halftone processing unit

Claims (12)

Image feature extraction processing that performs edge detection of multi-gradation image data and performs region expansion on the detection result to distinguish pixels of edge regions and non-edge regions of multi-tone image data;
White background determination processing for determining white background pixels and non-white background pixels of multi-tone image data;
A quantization process for quantizing multi-tone image data by an error diffusion method and outputting quantized data; and (1) for a pixel that is a white background pixel by the white background determination process, for the quantization process A quantization threshold value is fixed to a predetermined value. (2) For a pixel that is a non-white background pixel by the white background determination process and a pixel that is an edge area pixel by the image feature extraction process, the quantization process is performed. A quantization threshold for the pixel is fixed to the predetermined value, and (3) a pixel that is a non-white background pixel by the white background portion determination process and a non-edge portion pixel by the image feature extraction process is Processing to periodically oscillate the quantization threshold for quantization processing,
An image processing method comprising: Image feature extraction processing for detecting the edge degree of each pixel of the multi-tone image data by performing edge detection of the multi-tone image data and performing area expansion on the detection result,
White background determination processing for determining white background pixels and non-white background pixels of multi-tone image data;
A quantization process for quantizing multi-tone image data by an error diffusion method and outputting quantized data; and (1) for a pixel that is a white background pixel by the white background determination process, for the quantization process The quantization threshold is fixed to a predetermined value, and (2) for pixels that are non-white background pixels by the white background determination process, the smaller the degree of edge detected by the image feature extraction process, the greater the vibration width. Processing to periodically oscillate the quantization threshold for quantization processing,
An image processing method comprising: Edge detection processing for multi-tone image data,
Region expansion processing for the result of the edge detection processing;
White background determination processing for determining white background pixels and non-white background pixels of multi-tone image data;
Quantization processing for quantizing multi-tone image data by error diffusion method and outputting quantized data, and
(1) For a pixel that is a white background pixel by the white background determination process, a quantization threshold value for the quantization process for a pixel that is a pixel of an edge region by the edge detection process is fixed to a predetermined value. , Periodically oscillating the quantization threshold for the quantization processing for the pixels determined as the non-edge region pixels by the edge detection processing, and (2) the non-white background pixels by the white background determination processing Regarding the pixel, the quantization threshold for the quantization process for the pixel that has been made the pixel in the edge region by the region expansion process is fixed to the predetermined value, and the pixel that has been made the pixel in the non-edge region by the region expansion process A process of periodically oscillating a quantization threshold for the quantization process for
An image processing method comprising: Edge detection processing for multi-tone image data,
Region expansion processing for the result of the edge detection processing;
White background determination processing for determining white background pixels and non-white background pixels of multi-tone image data;
Quantization processing for quantizing multi-tone image data by error diffusion method and outputting quantized data, and
(1) For a pixel that is a white background pixel by the white background determination process, the quantization threshold of the quantization process is periodically changed with a larger vibration width as the edge degree detected by the edge detection process is smaller. (2) For a pixel that is a non-white background pixel by the white background determination process, the quantization threshold value of the quantization process is cycled with a larger vibration width as the edge degree detected by the area expansion process is smaller. And an image processing method characterized by comprising: 5. The image processing method according to claim 1, wherein in the white background portion determination process, the determination relating to the target pixel is performed based on an average level of multi-tone image data around the target pixel. A computer-readable storage medium having recorded thereon a program for causing a computer to execute each process for the image processing method according to claim 1. Image feature extraction means for performing edge detection of multi-gradation image data and discriminating pixels of edge regions and non-edge regions of multi-tone image data by performing region expansion on the detection results;
A white background determination means for determining a white background pixel and a non-white background pixel of the multi-tone image data;
Quantization processing means for quantizing multi-tone image data by error diffusion method and outputting quantized data,
(1) For a pixel that is determined as a white background pixel by the white background determination unit, a quantization threshold value of the quantization processing unit is fixed to a predetermined value, and (2) a non-white background pixel is detected by the white background determination unit. In addition, with respect to pixels that have been determined as edge region pixels by the image feature extraction means, the quantization threshold value of the quantization processing means is fixed to the predetermined value, and (3) non-white background pixels by the white background determination means And a quantization threshold control means for periodically oscillating the quantization threshold of the quantization processing means for the pixels that are non-edge pixels by the image feature extraction means,
An image processing apparatus comprising: Image feature extraction means for detecting the edge degree of each pixel of the multi-tone image data by performing edge detection of the multi-tone image data and performing area expansion on the detection result;
A white background determination means for determining a white background pixel and a non-white background pixel of the multi-tone image data;
Quantization processing means for quantizing multi-tone image data by error diffusion method and outputting quantized data,
(1) For a pixel that is determined as a white background pixel by the white background determination unit, a quantization threshold value of the quantization processing unit is fixed to a predetermined value, and (2) a non-white background pixel is detected by the white background determination unit. With respect to the pixel, a quantization threshold control unit that periodically oscillates the quantization threshold of the quantization processing unit with a large vibration width as the edge degree detected by the image feature extraction unit decreases.
An image processing apparatus comprising: The quantization threshold value control means includes
A quantization threshold generating means for generating a value that periodically vibrates with a large vibration width as the edge degree detected by the image feature extracting means decreases;
A pixel generated as a non-white background pixel by the white background determination unit uses a value generated by the quantization threshold generation unit as a quantization threshold, and a pixel determined as a white background pixel by the white background determination unit is predetermined. 9. The image processing apparatus according to claim 8, further comprising means for supplying the value as a quantization threshold to the quantization processing means. Edge detection means for multi-tone image data;
Area expansion processing means for performing area expansion processing on the detection result by the edge detection means;
A white background determination means for determining a white background pixel and a non-white background pixel of the multi-tone image data;
Quantization processing means for quantizing multi-tone image data by error diffusion method and outputting quantized data, and
(1) For a pixel that has been determined to be a white background pixel by the white background determination unit, the quantization threshold of the quantization processing unit for the pixel that has been determined to be an edge region pixel by the edge detection unit is fixed to a predetermined value. And periodically oscillating the quantization threshold value of the quantization processing means for the pixels determined as non-edge region pixels by the edge detection means, and (2) the non-white background pixels by the white background determination means For the pixel that has been processed, the quantization threshold value of the quantization processing unit for the pixel that has been determined as the pixel of the edge region by the region expansion processing unit is fixed to the predetermined value, and the non-edge region of the non-edge region is fixed by the region expansion processing unit. Quantization of the quantization processing means for a pixel Quantization threshold control means for periodically oscillating the threshold,
An image processing apparatus comprising: Edge detection means for performing edge detection processing on multi-tone image data;
Area expansion processing means for performing area expansion processing on the detection result of the edge detection means;
A white background determination means for determining a white background pixel and a non-white background pixel of the multi-tone image data;
Quantization processing means for quantizing multi-tone image data by error diffusion method and outputting quantized data, and
(1) For a pixel that is determined as a white background pixel by the white background portion determination unit, the quantization threshold value of the quantization processing unit is periodically changed with a larger vibration width as the edge degree detected by the edge detection unit is smaller. (2) For a pixel that has been determined as a non-white background pixel by the white background determination unit, the quantization processing unit performs quantization with a larger vibration width as the edge degree detected by the region expansion processing unit decreases. Quantization threshold control means for periodically oscillating the threshold,
An image processing apparatus comprising: The quantization threshold value control means includes
The pixel detected as the white background pixel by the white background determination means is the edge degree detected by the edge detection means, and the pixel expansion processing means is the pixel determined as the non-white background pixel by the white background determination means. Selection means for respectively selecting the edge degrees detected by
The quantization threshold of the quantization processor is periodically oscillated with a larger amplitude as the edge degree selected by the selector is smaller, and the quantization threshold of the quantization processor when the edge degree is the maximum 12. The image processing apparatus according to claim 11, comprising quantization threshold value generating means for fixing the value to a predetermined value.

Images