JP2006191458A - Image processor and image-processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor and an image-processing method capable of enhancing the sharpness while establishing the granularity and the resolution of an output image at the same time. <P>SOLUTION: The image processor for determining a ratio of dot-on pixels in accordance with a density value of multi-value image data and outputting the ratio in a form of small-value image data, includes: a shape recognition means 101 for recognizing the shape according to pattern recognition applied to the multi-value image data; an output order determining means 104 for determining dot output order on the basis of the recognized shape and image data around a target pixel; a quantization means 108 for applying quantization to the received multi-value image data into the small-value image data; and a dot output position control means 111 for controlling a dot-ON pixel position in accordance with a density value of the multi-value image data, and the dot output position control means arranges the dot-on pixels on the basis of the determined dot output order. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image processing apparatus and an image processing method applied to a laser printer, a digital copying machine, a color laser printer, a digital color copying machine, and the like, and in particular, performs an image quality improvement process for an output image of rasterized digital image data. The present invention relates to an image processing apparatus and an image processing method.

  Conventionally, in order to obtain an output image in which both the graininess and resolution of an image are simultaneously achieved, the following techniques are disclosed.

  In Patent Document 1, halftones are expressed by determining the ratio of ON and OFF pixels according to the density value of the original image, and differ according to the degree of change in density value in the halftone image mesh. A technique for redistributing ON pixels in a predetermined number of meshes using an image processing method is disclosed.

  Specifically, when the density gradient in the mesh is small, the dot concentration pattern order, when the density gradient is medium, when combining the dot concentration pattern considering the density value of each pixel, and when the density gradient is large The ON pixels in the mesh are redistributed in order of density. Furthermore, when the density gradient is small or medium, it is checked whether there is a place where the density gradient is high at any one pixel, and if there is no change in the number of ON pixels in the mesh, the highest density in the mesh The distribution is performed such that the pixels are turned on and the lowest density pixel is turned off.

Patent Documents 2 and 3 provide high-speed error diffusion processing capable of expressing gradation while maintaining resolution in a high-density writing type electrophotographic printer, and image graininess and resolution. A technique relating to an image processing method for improving the image quality is disclosed by the present applicant. Specifically, low-resolution input multivalued image data can be processed at high speed by performing multivalued error diffusion processing at low resolution using a dither threshold value corresponding to the edge degree of the image. It is converted into high resolution binary image data.
Japanese Patent Laid-Open No. 3-119861 JP 2002-185785 A JP 2002-344742 A

  In the image processing apparatus of Patent Document 1, when there is a place where the density gradient has jumped out of any one pixel in order to improve local resolution even when the density gradient is small or medium, Rather than changing the mesh to the ON pixel allocation method in order of density, the number of ON pixels in the mesh is not changed as described above, the highest density pixel in the mesh is turned ON, and the lowest density is set. The resolution is improved without deteriorating the graininess by the method of turning off the pixels.

  However, in the method of Patent Document 1, since halftones are expressed by ON / OFF of each pixel in the mesh, that is, binary output, the number of gradations to be obtained is limited, and the number of gradations should be increased. When the mesh is enlarged, the sharpness deteriorates, so that there is a problem that the image quality is not satisfactory.

  Therefore, in Patent Documents 2 and 3, an error diffusion method capable of expressing gradation while maintaining resolution is used, and a dither threshold corresponding to the degree of edge of the image is used for input multilevel image data. By performing error diffusion processing, the image quality of graininess and resolution is improved with a sufficient number of gradations. In Patent Documents 2 and 3, high-speed processing is possible because low-resolution input multilevel image data is converted into high-resolution binary image data by performing multilevel error diffusion at low resolution.

  However, in Patent Documents 2 and 3, when the edge degree is large, the dither threshold value is set as a fixed threshold value, and the dot output position to the high resolution pixel is in a predetermined arrangement order. There is a problem that the resolution is not obtained.

  The present invention has been made in view of the above points, and it is an object of the present invention to provide an image processing apparatus and an image processing method capable of improving sharpness while simultaneously achieving both the granularity and resolution of an output image. And

  In order to solve the above-described problem, the present invention determines the dot-on pixel ratio according to the density value of the input multi-value image data, and uses the low-value image data having a multi-value number smaller than the multi-value image data. An image processing apparatus that outputs a shape recognition unit that recognizes a shape by pattern recognition on input multi-valued image data, and a shape recognized by the shape recognition unit and image data around a target pixel. Based on the dot output order determining means for determining the dot output order, the quantizing means for quantizing the input multi-value image data into the low-value image data, and the density value of the input multi-value image data Dot output position control means for controlling the dot-on pixel position, the dot output position control means based on the dot output order determined by the dot output order determination means. And wherein placing the Toon pixel.

  The present invention is also an image processing apparatus for outputting input low-resolution multilevel image data as high-resolution low-value image data having a multi-value number smaller than that of the multi-value image data. Shape recognition means for recognizing shape by pattern recognition for multi-valued image data, and dot output order determination for determining dot output order based on the shape recognized by the shape recognition means and image data around the target pixel Means, quantization means for quantizing the inputted multi-value image data into low-value image data in units of low resolution, and the low-resolution low-value image data quantized by the quantization means Dot output position control means for controlling the dot-on pixel position by converting it into the number of dot-on pixels of resolution low-value image data, the dot output position control means determining the dot output order Based on the dot output sequence determined by stage, characterized by arranging the dot-on pixels of the small value image data of the high resolution.

  The present invention is also an image processing apparatus for outputting input low-resolution multilevel image data as high-resolution low-value image data having a multi-value number smaller than that of the multi-value image data. Shape recognition means for recognizing shape by pattern recognition for multi-valued image data, and dot output order determination for determining dot output order based on the shape recognized by the shape recognition means and image data around the target pixel Means, quantization means for quantizing the inputted multi-value image data into low-value image data by multi-value error diffusion in units of low resolution, and the low-resolution low value quantized by the quantization means Dot output position control means for converting the image data into the number of dot-on pixels of high-resolution low-value image data and controlling the dot-on pixel position, the dot output position control means, Based on the dot output order determined by Tsu preparative output order determining means, characterized by arranging the dot-on pixels of the small value image data of the high resolution.

  In addition, the present invention provides an image processing method for determining a dot-on pixel ratio according to a density value of input multi-valued image data, and outputting with less value image data having a smaller number of multi-values than the multi-value image data. The shape of the input multi-value image data is recognized by pattern recognition, the dot output order is determined based on the recognized shape and the image data around the target pixel, and the input multi-value image Quantizing the data into low-value image data, controlling the dot-on pixel position according to the density value of the input multi-value image data, and arranging the dot-on pixel based on the determined dot output order Features.

  The present invention is also an image processing method for outputting input low-resolution multi-value image data as high-resolution low-value image data having a multi-value number smaller than that of the multi-value image data. Recognize the shape of the multi-valued image data by pattern recognition, determine the dot output order based on the recognized shape and the image data around the pixel of interest, and input multi-valued image data in low-resolution pixel units Quantized into low-value image data in the above, and the quantized low-resolution low-value image data is converted into the number of dot-on pixels in the high-resolution low-value image data, and the dot-on pixel position is controlled and determined. The dot-on pixels of the high-resolution low-value image data are arranged based on the dot output order.

  The present invention is also an image processing method for outputting input low-resolution multi-value image data as high-resolution low-value image data having a multi-value number smaller than that of the multi-value image data. Recognize the shape of the multi-valued image data by pattern recognition, determine the dot output order based on the recognized shape and the image data around the pixel of interest, and input multi-valued image data in low-resolution pixel units Quantize to low-value image data by multi-value error diffusion and convert the quantized low-resolution low-value image data to the number of dots-on pixels in high-resolution low-value image data to control the dot-on pixel position. The dot-on pixels of the high-resolution low-value image data are arranged based on the determined dot output order.

  In the present invention, the shape of the input multi-valued image data is recognized by pattern recognition, and the arrangement of dot-on pixels is determined based on the recognition result and the image data around the target pixel. In this edge portion, it is possible to arrange a dot-on image faithful to the shape of the input multi-value image data.

  According to the present invention, it is possible to provide an image processing apparatus and an image processing method capable of improving sharpness while simultaneously satisfying both the graininess and resolution of an output image.

  Next, the best mode for carrying out the present invention will be described based on the following embodiments with reference to the drawings. In the present embodiment, an example will be described in which the input gradation value and the output gradation value take integer values of 0 or more and 255 or less. The input gradation value and the output gradation value “0” represent the lowest density. The input gradation value and the output gradation value “255” represent the highest density.

  In the first embodiment of the present invention, jaggy reduction processing using pattern matching is performed on 600 dpi, 256-gradation multilevel input image data, and quinary error diffusion processing is performed on the output result of the jaggy reduction processing. An example of generating a 1200 dpi binary image based on the result of image shape recognition and the edge level is shown for the quantization result of the quinary error diffusion processing.

  FIG. 1 is a block diagram showing a first embodiment of an image processing apparatus according to the present invention. The image processing apparatus of FIG. 1 has a configuration having a jaggy reduction processing unit and a halftone processing unit. The jaggy reduction processing unit includes a shape recognition unit 101, a correction value reading unit 102, a data correction unit 103, a template ROM 1011 and a correction value ROM 1021.

  The halftone processing unit includes a dot output order determination unit 104, an edge level calculation unit 105, a γ correction unit 106, an adder 107, a quantization unit 108, a subtractor 109, an error calculation unit 110, and a dot output position control unit 111. It is the composition which has.

  First, the shape recognizing unit 101 recognizes a shape by pattern recognition on input image data having 600 dpi and 256 gradation multilevel. When pattern recognition is performed, a plurality of types of templates as shown in FIGS. 2A and 2B are prepared in advance and stored in the template ROM 1011. FIG. 2 is an image diagram showing an example of a template.

  The shape recognizing unit 101 determines whether or not the image has a specific shape by matching the peripheral data window centering on the target pixel to be processed with the template for each small piece and detecting a match between the two. When a match with the template is detected and the shape of the image is specified, the shape recognition unit 101 passes the template information of the matched template to the correction value reading unit 102.

  Based on the template information received from the shape recognition unit 101, the correction value reading unit 102 reads correction value data used to replace the pixel of interest with correction dots in the next data correction unit 103 from the correction value ROM 1021. FIG. 3 is an image diagram illustrating an example in which correction value data is stored in association with template information. For example, when the template information received from the shape recognition unit 101 is “Template No. 1”, the correction value reading unit 102 reads the corresponding correction value data “160” from the correction value ROM 1021 and passes it to the data correction unit 103. . The data correction unit 103 replaces the target pixel of the input image with the correction value data read by the correction value reading unit 102.

  With the above processing, for example, even when the input image is an image close to a horizontal line that easily generates jaggy as shown in FIG. 4, the jaggy reduction processing unit in FIG. An image as shown in FIG. 5 is obtained as an output of the data correction unit 103 by performing correction by the data correction unit 103 using correction value data obtained from the matched template information for each neighboring pixel.

  If the image shown in FIG. 5 is used, the halftone processing unit converts the data into 1200 dpi binary data, so that jaggy can be reduced when the image is converted into a 1200 dpi binary image as shown in FIG.

  On the other hand, when the input image as shown in FIG. 4 is converted into 1200 dpi binary data without performing the jaggy reduction processing by the shape recognition unit 101, the correction value reading unit 102, and the data correction unit 103 described above, intermediate Even when the tone processing unit converts the data into 1200 dpi binary data, an image as shown in FIG. 7 is obtained, and an output image with noticeable jaggy is obtained. When the image of FIG. 6 and the image of FIG. 7 are output using an actual 1200 dpi laser printer and compared with each other, the image quality is obvious.

  If the peripheral data window does not match any of the prepared templates, the shape recognition unit 101 does not correct the target pixel in the data correction unit 103 and outputs the input image data through. The template information detected by the shape recognizing unit 101 is also passed to the dot output order determining unit 104 of the halftone processing unit, and is also used for determining the dot output order when performing high-resolution pixel dot arrangement. The dot output order determination unit 104 uses the template information detected by the shape recognition unit 101 and the image data around the target pixel including the target pixel whose data has been corrected by the data correction unit 103 to perform next dot output position control. The dot output order for use in the unit 111 is determined. Details of the dot output order determination unit 104 and the dot output position control unit 111 will be described later.

  The edge level calculation unit 105 calculates a plurality of edge levels for the image whose data has been corrected by the data correction unit 103. In this embodiment, there are two stages of edge level 0 or edge level 1. However, it is assumed that edge level 1 has a higher degree of image edge than edge level 0.

  FIG. 8 is a flowchart showing the processing of the edge level calculation unit. First, in step S1, the edge level calculation unit 105 calculates the edge amount of each pixel of 600 dpi using the edge extraction filter shown in FIG.

  The edge extraction filter of FIG. 9 extracts edge amounts of FIG. 9 (a) as vertical lines, FIG. 9 (b) as horizontal lines, and FIGS. 9 (c) and 9 (d) as diagonal lines. Specifically, the edge level calculation unit 105 multiplies the four edge amounts calculated in FIGS. 9A to 9D by corresponding coefficients. For example, the coefficient in FIGS. 9A and 9B is 2. The coefficient in FIGS. 9C and 9D is 1. The edge level calculation unit 105 sets the maximum value of the edge amount of each edge extraction filter multiplied by the coefficient as the edge amount of the pixel.

  Next, in step S2, the edge level calculation unit 105 determines whether or not the target pixel belongs to a white background area. When there are a predetermined number (for example, 6 pixels) or more of white pixels in a predetermined area (for example, 5 × 5 area) centered on the target pixel, the edge level calculation unit 105 determines that the target pixel belongs to the white background area. Here, the white pixel is preferably a pixel taking noise into consideration, for example, a pixel whose data (density) is 5 or less.

  Next, the edge level calculation unit 105 switches the edge amount threshold based on the result of the white background determination, and quantizes the calculated edge amount into a plurality of levels. If the pixel of interest belongs to the white background area (YES in S2), the edge level calculation unit 105 proceeds to step S3, and sets the edge amount (quantization) threshold value to a lower value compared to the case where it does not belong to the white background area. If the pixel of interest does not belong to the white background area (NO in S2), the edge level calculation unit 105 proceeds to step S4, and sets the edge amount (quantization) threshold value to a higher value than when it belongs to the white background area.

  The reason for such switching is to make it easy to determine even a character with a relatively low edge amount such as a low-density character as an edge (since there are many white pixels around the character region, a low threshold is easily selected. ). Then, the edge level calculation unit 105 proceeds to step S5 following step S3 or S4, and quantizes the calculated edge amount to a plurality of levels according to the edge amount threshold value switched as necessary in step S3 or S4. .

  In step S6, the edge level calculation unit 105 determines whether the target pixel is a black pixel. The edge level calculation unit 105 determines that the pixel of interest is a black pixel if the data of the pixel of interest is equal to or greater than a predetermined value (YES in S6), and proceeds to step S7 to set the edge level of the pixel of interest to the maximum level (in this embodiment, = 1). This is to make it easy to determine an edge even in a high density halftone dot portion having a relatively small edge amount. Note that the edge level calculation unit 105 determines that the pixel of interest is not a black pixel if the data of the pixel of interest is not equal to or greater than a predetermined value (NO in S6), and ends the process. Needless to say, the edge extraction filter may be other than the example shown in FIG.

  The γ correction unit 106 performs γ correction processing using the γ table. Although not shown in the present embodiment, the two γ tables may be switched according to the edge level calculated by the edge level calculation unit 105 described above.

  The quantization unit 108 performs multilevel error diffusion processing at 600 dpi on the output data from the γ correction unit 106 and quantizes the data to five values. The error diffusion process is realized by the adder 107, the quantization unit 108, the subtractor 109, and the error calculation unit 110.

  The adder 107 adds the quantization error of the processed pixel to the multi-value image data. The quantization unit 108 outputs the output data of the adder 107 by four quantization thresholds Thr1, Thr2, Thr3, and Thr4 (in this embodiment, Thr1 = 32, Thr2 = 96, Thr3 = 160, Thr4 = 224). Perform five-value quantization. The subtractor 109 calculates a quantization error from the input and output of the quantization unit 108. The error calculation unit 110 calculates an error amount to be added to the pixel to be processed next from the quantization error calculated by the subtractor 109 according to a predetermined error matrix, and supplies the error amount to the adder 107.

  The relationship between the input value and the output value of the quantizing unit 108 is as follows: output value = 0 when the input value <32, output value = 64 when 32 ≦ input value <96, and output value = 128 when 96 ≦ input value <160. , 160 ≦ input value <224, output value = 192, and 224 ≦ input value, output value = 255.

  Although the present embodiment shows an example in which the values of the quantization thresholds Thr1 to Thr4 are set at equal intervals, this is not particularly limited, and there is no problem even if they are set unevenly. In this embodiment, multilevel error diffusion is adopted for quantization. However, this is not particularly limited, and there is no problem even with simple quinary processing that does not feed back quantization error.

  Details of the dot output order determination unit 104 will be described. FIG. 10 is a flowchart showing the process of the dot output order determination unit. First, in step S11, the dot output order determination unit 104 determines whether or not the template information detected by the shape recognition unit 101 exists. The dot output order determination unit 104 uses different methods depending on whether the template information is present or not, and the dot at 1200 dpi 2 × 2 pixels as shown in FIG. 6 corresponding to the target pixel at 600 dpi pixels. Determine the output order.

  First, when there is template information from the shape recognition unit 101 (YES in S11), the dot output order determination unit 104 proceeds to step S12, and the dot output order corresponding to the template information stored in advance in the template ROM 1011 or the like. Is read.

  FIG. 12 is an image diagram illustrating an example in which the dot output order is stored in association with the template information. The pixel number “A” is the upper left pixel of 2 × 2 pixels of 1200 dpi in FIG. 11, the pixel number “B” is the upper right pixel, the pixel number “C” is the lower left pixel, and the pixel number “D” is the lower right pixel. In this image diagram, pixel numbers are arranged in the order in which dots are placed from the left.

  For example, when the template information received from the shape recognizing unit 101 is “Template No. 1”, ABCD is described in the output order column, so the corresponding dot output order is the upper left pixel and upper right pixel in FIG. , Lower left pixel, lower right pixel. When the template information received from the shape recognition unit 101 is “Template No. 2”, since DCBA is described in the output order column, the corresponding dot output order is the lower right pixel in FIG. Pixel, upper right pixel, upper left pixel.

  On the other hand, when the template information from the shape recognition unit 101 does not exist (NO in S11), the dot output order determination unit 104 proceeds to step S13 and sets the dot generation order based on the density distribution of the pixel-of-interest peripheral data. decide. Specifically, the dot output order determination unit 104 has three areas in the upper left, upper right, lower left, and lower right areas around the target pixel as shown in FIGS. 13A to 13D. Add pixels. In step S14, the dot output order determination unit 104 compares the addition results shown in FIGS. 13A to 13D and arranges them in descending order.

  In step S15, the dot output order determination unit 104 compares the comparison results for the four regions of the upper left, upper right, lower left, and lower right with respect to the upper left, upper right, lower left, and lower right pixels in the 1200 dpi pixel of FIG. Consider it a result. The upper left area in FIG. 13A is replaced with an upper left pixel A in FIG. 11 having 2 × 2 pixels of 1200 dpi. The upper right area in FIG. 13B is replaced with the upper right pixel B in FIG. The lower left area in FIG. 13C is replaced with the lower left pixel C in FIG. The lower right region in FIG. 13D is replaced with the lower right pixel D in FIG. The dot output order determination unit 104 sets the pixels A, B, C, and D arranged in descending order as the dot output order as they are.

  For example, if the addition results in FIGS. 13A to 13D are in the order of FIG. 13C, FIG. 13A, FIG. 13D, and FIG. Becomes CADB. In this case, the dot output order of 2 × 2 pixels of 1200 dpi in FIG. 11 is the lower left pixel C, the upper left pixel A, the lower right pixel D, and the upper right pixel B.

  When there is template information in this way, by determining the dot output order based on the template information, it is possible to reliably place dots that are faithful to the shape of the input image regardless of the state of the surrounding pixels. It leads to improvement of sex.

  For example, when the data correction unit 103 performs data correction on the input image as shown in FIG. 4 and the result is an image as shown in FIG. 5, the ideal dot output order for the pixel “S” is DCBA. Is. However, when the dot output order is determined by the above method that does not use template information, the determined dot output order is CDBA, resulting in different results. If the input image is a halftone line drawing, this order difference greatly affects the sharpness. In this embodiment, the dot output order when there is no template information is determined by the above method, but it is not limited to this method.

  The dot output order thus obtained is a combination of 4 pixels of 2 × 2 in total. Progressing to step S16 following step S12 or S15, the dot output order determination unit 104 encodes the determined output order result and passes the result to the dot output position control unit 111.

  Next, the dot output position control unit 111 will be described. The dot output position control unit 111 converts 600 dpi 5-value data obtained by error diffusion processing into 1200 dpi 2 × 2 pixel dot on / off (binary image data).

  The dot output position control unit 111 first calculates the number of output dots. That is, the dot output position control unit 111 determines the number of dots-on pixels (number of dots) in a 1200 dpi 2 × 2 pixel (FIG. 11) corresponding to each pixel on 600 dpi multi-valued image data.

  Specifically, the dot output position control unit 111 sets 0 when the quantized data value (5 values) of the quantized data from the quantizing unit 108 is 0, and sets 1 when the quantized data value is 64. The number of dots is calculated as 2 when the quantized data value is 128, 3 when the quantized data value is 192, and 4 when the quantized data value is 255.

  Next, based on the calculated number of dots, the dot output position control unit 111 determines the arrangement of dots in a 1200 × 2 2 × 2 pixel corresponding to each pixel. In this embodiment, a different method is adopted as the dot arrangement method based on the edge level information from the edge level calculation unit 105.

  When the edge level = 1 (the edge degree is large), the dot arrangement based on the output order from the dot output order determination unit 104 is performed. On the other hand, when the edge level = 0 (the edge degree is small), the dot arrangement based on the output order from the dot output order determination unit 104 is not performed, and as shown in FIG. × 2 pixels are arranged in a fixed order.

  FIG. 14 is an image diagram illustrating an example in which dot arrangement based on the output order from the dot output order determination unit is not performed. In the image diagram of FIG. 14, the upper left pixel of 2 × 2 pixels of 1200 dpi is the pixel number “1”, the upper right pixel is the pixel number “2”, the lower left pixel is the pixel number “3”, and the lower right pixel is the pixel number “4”. Yes.

  That is, when the number of dots is 1, a dot is arranged at the pixel with the pixel number “1” in FIG. When the number of dots is 2, in addition to the pixel with the pixel number “1”, dots are also arranged on the pixel with the pixel number “2”. When the number of dots is 3, in addition to the pixels with the pixel numbers “1” and “2”, dots are also arranged on the pixel with the pixel number “3”. When the number of dots is 4, dots are arranged on all pixels.

  In this way, in the portion where the degree of edge is large, the effect of finely controlling smaller dots can be obtained by optimizing the dot arrangement with 1200 dpi pixels based on the shape recognition result, and the sharpness is improved. Can be expected. On the other hand, when a method similar to the case where the edge degree is large is adopted in a portion where the edge degree is small, a slightly noisy image is obtained. Therefore, by arranging the dots in a fixed order in units of 1200 dpi, a multi-value of 600 dpi is obtained. The image quality is equivalent to error diffusion. However, even in the case of an image pattern that becomes an isolated dot with 600 dpi multi-value, binary pixels are often arranged adjacent to two dots at 1200 dpi, so that the granularity in an output device such as a laser printer with poor isolated dot reproducibility. This may be advantageous.

  In this embodiment, the dot arrangement method is switched between the case where the edge degree is large and the case where the edge degree is small, but it is not always necessary to perform the switching, and in any case, the output order from the dot output order determination unit 104 The effect of improving the sharpness is not lost even if the dot arrangement based on the above is performed.

  Further, in this embodiment, an example in which binary image data is output has been shown. However, output dots are not limited to binary image data, and there are fewer levels than input multivalued image data such as ternary and quaternary values. If it is small value data of the key value, it can be similarly implemented. For example, when the output is ternary data, the quantization number is increased to, for example, 9-valued, and when the result is converted into the number of dot-on pixels of high-resolution pixels, the quantization result is 0 and the number of dot-ons = 0, quantization result is 1 or 2, dot on number = 1, quantization result is 3 or 4, dot on number = 2, quantization result is 5 or 6, dot on number = 3, quantization result is 7 or 8, dot on number = 4 In other words, the number of dot-ons including the intermediate dots of the ternary data may be determined to control the pixel position where the dots are turned on.

  In the second embodiment of the present invention, jaggy reduction processing using pattern matching is performed on 600 dpi, 256-gradation multi-valued input image data, and a 5-value using a dither threshold is used for the output result of the jaggy reduction processing. An example will be shown in which error diffusion processing is performed and a 1200 dpi binary image is generated based on the result of image shape recognition and the edge level for the quantization result of the quinary error diffusion processing. Here, the dither threshold is also switched based on the edge level.

  FIG. 15 is a block diagram showing a second embodiment of the image processing apparatus according to the present invention. The image processing apparatus in FIG. 15 is assigned the same number as the image processing apparatus in FIG. 1 for performing the same processing as the image processing apparatus in FIG. 1 used in the description of the first embodiment.

  First, the shape recognizing unit 101 recognizes a shape by pattern recognition for 600 dpi, 256-gradation multi-valued input image data in the same manner as in the first embodiment, and sends the matched template information to the correction value reading unit 102. hand over. The correction value reading unit 102 reads correction value data from the correction value ROM 1021 based on the template information received from the shape recognition unit 101 and passes the correction value data to the data correction unit 103. The data correction unit 103 replaces the target pixel of the input image with the correction value data read by the correction value reading unit 102.

  The template information detected by the shape recognition unit 101 is also passed to the dot output order determination unit 104 as in the first embodiment. The dot output order determination unit 104 determines the dot output order using the template information and the image data from the data correction unit 103. Since the dot output order determination method here is the same as that of the first embodiment, the description thereof is omitted. Further, the operations of the edge level calculation unit 105 and the γ correction unit 106 are the same as those in the first embodiment, and thus description thereof is omitted.

  From here, the part of the processing content in which operation | movement differs from Example 1 is demonstrated. First, when the edge level = 0 output from the edge level calculation unit 105, the dither threshold selection unit 212 selects four dither threshold matrices (a) to (d) shown in FIG. A position value corresponding to the target pixel position is set as a threshold value (thr1 to thr4). On the other hand, when the edge level = 1, the dither threshold selection unit 212 fixes the thresholds (thr1 to thr4) regardless of the pixel position, and for example, thresholds similar to those used in the first embodiment (Thr1 = 32, Thr2 = 96). , Thr3 = 160, Thr4 = 224).

  Next, the quantization unit 208 performs multilevel error diffusion processing at 600 dpi on the output data from the γ correction unit 106 using the threshold value determined by the dither threshold value selection unit 212, and quantizes it to five values. The error diffusion process is realized by the adder 107, the quantization unit 208, the subtractor 109, and the error calculation unit 110.

  The adder 107 adds the quantization error of the processed pixel to the multi-value image data. The quantization unit 208 performs five-value quantization on the output data of the adder 107 by four quantization thresholds Thr1, Thr2, Thr3, Thr4 (Thr1 <Thr2 <Thr3 <Thr4). The subtractor 109 calculates a quantization error from the input and output of the quantization unit 208. The error calculation unit 110 calculates an error amount to be added to the pixel to be processed next from the quantization error calculated by the subtractor 109 according to a predetermined error matrix, and supplies the error amount to the adder 107.

  The relationship between the input value and the output value of the quantization unit 208 is as follows: output value = 0 when the input value <Thr1, output value = 64 when Thr1 ≦ input value <Thr2, and output value = 128 when Thr2 ≦ input value <Thr3. When Thr3 ≦ input value <Thr4, output value = 192, and when Thr4 ≦ input value, output value = 255.

  The 6 × 6 (3 × 3 at 600 dpi) dither threshold matrix shown in FIG. 17 is a spiral array of thresholds that increase in step width 3 from 74 to 179 in ascending order of value, and 200 lines at 1200 dpi. This is a dot concentration type that forms halftone dots. In FIG. 17, four threshold values inside the solid grid correspond to one pixel of 600 dpi. As shown in FIG. 17, the threshold values (74, 77, 80, 83) from the smallest to the fourth in this dither threshold matrix are arranged at different pixel positions of 600 dpi.

  That is, if the upper left 2 × 2 pixels (110, 107, 113, and 80 pixels) in FIG. 17 are assigned, the lowest 80 is assigned to the upper left pixel in FIG. (B) The next lowest 110 is assigned to FIG. 16C, and the highest 113 is assigned to FIG. 16D.

  The dither threshold selection unit 212 outputs four thresholds corresponding to one pixel of 600 dpi in the dither threshold matrix as quantization thresholds corresponding to the pixel. For example, in the upper left pixel position, the minimum 80 of the four threshold values (80, 107, 110, 113) at that position is the quantization threshold Thr1, the next smaller 107 is the childization threshold Thr2, and the next smaller 110 is the quantum. The quantization threshold Thr3 and the largest 113 are output as the quantization threshold Thr4.

  The 3 × 3 dither threshold matrix shown in FIG. 16 is obtained by rearranging the thresholds in the dither threshold matrix shown in FIG. 17 at a pixel position of 600 dpi for each quantization threshold. FIG. 16A corresponds to the quantization threshold Thr1. FIG. 16B corresponds to the quantization threshold Thr2. FIG. 16C corresponds to the quantization threshold Thr3. FIG. 16D corresponds to the quantization threshold Thr4. Actually, four 3 × 3 dither threshold matrixes shown in FIG. 16 are stored in advance in the dither threshold ROM 2121 so as to be called as the threshold matrix when the edge level = 0.

  Next, the dot output position control unit 211 will be described. As in the first embodiment, the dot output position control unit 211 converts 600 dpi 5-value data obtained by error diffusion processing into 1200 dpi 2 × 2 pixel dot on / off (binary image data).

  The dot output position control unit 211 first calculates the number of output dots by the same method as in the first embodiment. Next, the dot output position control unit 211 determines the dot output position within the 1200 dpi 2 × 2 pixel corresponding to each pixel based on the calculated number of dots. In this embodiment, a different method is adopted as the dot arrangement method based on the edge level information from the edge level calculation unit 105.

  When the edge level = 1 (the edge degree is large), dot placement based on the output order from the dot output order determination unit 104 is performed in the same manner as in the first embodiment. On the other hand, when the edge level = 0 (the edge degree is small), the number of dots determined in accordance with the corresponding position on the dither threshold matrix (FIG. 16) of the processing target pixel is within 2 × 2 pixels of 200 dpi. Decide how to place them.

  The case where the edge level = 0 is described in detail. For example, when the pixel at the upper left pixel position in the dither threshold matrix shown in FIG. 17 is to be processed, a dot is first arranged at the position having the smallest value among the 2 × 2 threshold values at the upper left, and then The second dot is arranged at the position of the large threshold, the third dot is arranged at the position of the next largest threshold, and the last dot is arranged at the position of the largest threshold. In this case, if the number of dots is 2, as shown in FIG. 18 (a), dots are output at the right two pixel positions (thresholds 80 and 107) in 2 × 2 pixels (the two pixels are dot-on). Pixel). If the number of dots is 1, a dot is output at the lower right pixel position. If the number of dots is 3, dots are output to the upper left pixel position in addition to the right two pixel positions.

  Similarly, when the pixel on the right is to be processed, if the number of dots is 2, dots are output to the lower two pixel positions (threshold values 77 and 98) as shown in FIG. become. In this embodiment, halftone dots are easily formed by changing the dot output order (arrangement order) according to the corresponding position of the pixel to be processed on the dither threshold matrix.

  In this manner, halftone dots are formed using a dither threshold in a relatively flat portion of an image with an edge level = 0, so that an image with excellent graininess and stability can be formed. On the other hand, in a region with a large edge degree of edge level = 1, the effect of finely controlling smaller dots can be obtained by optimizing the dot arrangement with 1200 dpi pixels based on the shape recognition result, as in the first embodiment. As a result, improvement in sharpness can be expected.

  In the third embodiment of the present invention, the jaggy reduction process is not performed, the quinary error diffusion process using the dither threshold is performed on the input image data, and the shape recognition of the image is performed on the quantization result of the quinary error diffusion process. An example of generating a 1200 dpi binary image based on the result and the edge level is shown. Here, the dither threshold is also switched based on the edge level.

  FIG. 19 is a block diagram showing a third embodiment of the image processing apparatus according to the present invention. The image processing apparatus in FIG. 19 is assigned the same number as the image processing apparatus in FIG. 15 for performing the same processing as the image processing apparatus in FIG. 15 used in the description of the second embodiment.

  The image processing apparatus in FIG. 19 is different from the image processing apparatus in FIG. 15 in that the jaggy reduction processing unit is deleted. Specifically, in the image processing apparatus of FIG. 19, the correction value reading unit 102 and the data correction unit 103 that constitute the jaggy reduction processing unit are deleted, and the shape recognition unit 101 that constitutes the jaggy reduction processing unit is intermediate. It differs from the image processing apparatus of FIG. 15 in that the positional relationship is in parallel with the edge level calculation unit 105 of the tone processing unit.

  Even in the case of the configuration as shown in FIG. 19, the image processing apparatus is based on the output order from the dot output order determination unit 104 reflecting the shape recognition result by pattern recognition in the area with a large edge degree, as in the first and second embodiments. Therefore, the effect of improving the sharpness is maintained.

  The fourth embodiment of the present invention shows an example in which a binary image having an output of 600 dpi is generated for 600 dpi multi-value input image data. Again, the dot output position control method is switched according to the edge level. In this embodiment, since both input and output are 600 dpi, a dot position control process is performed in which pixels in a predetermined area are added once and a dot arrangement in the predetermined area is determined based on the addition result.

  FIG. 20 is a block diagram showing a fourth embodiment of the image processing apparatus according to the present invention. In the image processing apparatus of FIG. 20, the same number as that of the image processing apparatus of FIG. 19 is assigned to the same processing as that of FIG. 19 used in the description of the third embodiment. The image processing apparatus in FIG. 20 is different from the image processing apparatus in FIG. 19 in a dot output order determination unit 304, an in-area pixel addition unit 313, a quantization unit 308, and a dot output position control unit 311.

  When the predetermined area is 3 × 3 pixels, the in-area pixel addition unit 313 first adds the data values of 9 pixels included in the area. Then, the next quantization unit 308 quantizes the data into 10-value data. Here, nine quantization thresholds may be prepared at equal intervals. On the other hand, the dot output order determination unit 304 stores the template ROM 1011 or the like in advance when there is a matched template within a predetermined area as a result of pattern recognition similar to that in the third embodiment performed by the shape recognition unit 101. The dot output order corresponding to the template information is read, and when the template information does not exist, the order in the predetermined area is determined by the density distribution of the peripheral region as in the third embodiment. Then, the dot output position control unit 311 arranges the data that has been digitized by 600 dpi by the quantization unit 308 into 3 × 3 pixels within a predetermined area.

  First, the dot output position control unit 311 calculates the number of output dots by the same method as in the first embodiment. Next, the dot output position control unit 311 determines the dot output position. Also in this embodiment, a different dot arrangement method is adopted based on the edge level information from the edge level calculation unit 105.

  When edge level = 1 (high edge degree), dot placement based on the output order from the dot output order determination unit 304 is performed as in the third embodiment. On the other hand, when the edge level = 0 (edge degree is small), dot placement is performed according to the density pattern stored in the density pattern ROM 3111 in advance.

  As described above, in the relatively flat portion of the image with the edge level = 0, the dots are arranged according to the density pattern, so that an image having excellent graininess and stability can be formed. On the other hand, in a region with a large edge degree of edge level = 1, improvement of sharpness can be expected by optimizing the dot arrangement based on the shape recognition result as in the other embodiments.

  In the embodiment of the present invention, the edge level is calculated from the image data immediately before the γ correction processing by the γ correction unit 106, but it is not always necessary to use the image data at this position.

  As described above, according to the present invention, shape recognition by pattern recognition is performed on input image data, and the arrangement of output dots is determined based on the recognition result, so that particularly at the edge portion of an image. Therefore, it is possible to make the dot arrangement faithful to the shape of the input image data, and it is possible to improve the sharpness.

  In addition, in an image processing apparatus capable of high-resolution output, fine control of small dots is performed at the edge portion based on the shape recognition result without significant increase in processing time, and dot placement faithful to the shape of input image data is performed. Therefore, sharpness can be improved. Furthermore, in an image processing apparatus capable of high-resolution output, by forming halftone dots using a dither threshold value in a flat portion, an image having excellent graininess and stability can be formed. Therefore, according to the present invention, it is possible to obtain an output image in which the granularity and resolution of the image are compatible with each other with high quality.

  The present invention is not limited to the specifically disclosed embodiments, and various modifications and changes can be made without departing from the scope of the claims.

1 is a configuration diagram illustrating a first embodiment of an image processing apparatus according to the present invention. It is an image figure showing an example of a template. It is an image figure showing the example by which correction value data is matched and stored with template information. It is an image figure showing an example of an image which is easy to generate jaggy. It is an image figure showing an example of the image which performed the jaggy reduction process. It is an image figure of an example of 1200 dpi binary data which converted the image which performed the jaggy reduction process. It is an image figure of an example of the data of 1200 dpi binary which converted the image which is easy to generate | occur | produce jaggies. It is a flowchart showing the process of an edge level calculation part. It is an image figure of an example of an edge extraction filter. It is a flowchart showing the process of a dot output order determination part. It is an image figure of an example showing a 2 * 2 pixel of 1200 dpi. It is an image figure of an example showing the example in which the dot output order is stored in association with the template information. It is an image figure of an example showing the four area | regions of the upper left of the attention pixel periphery, an upper right, lower left, and lower right. It is an image figure of an example showing the example which does not perform dot arrangement based on the output order from a dot output order determination part. It is a block diagram which shows 2nd Example of the image processing apparatus by this invention. It is an image figure of an example of a 3x3 dither threshold value matrix. FIG. 6 is an image diagram of an example of a 6 × 6 dither threshold matrix. It is an image figure showing the process which arrange | positions a dot in an order from a position with a small value within the 2 * 2 threshold value of a dither threshold value matrix. It is a block diagram which shows 3rd Example of the image processing apparatus by this invention. It is a block diagram which shows 4th Example of the image processing apparatus by this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Shape recognition part 102 Correction value reading part 103 Data correction part 104,304 Dot output order determination part 105 Edge level calculation part 106 γ correction part 107 Adder 108,208,308 Quantization part 109 Subtractor 110 Error calculation part 111, 211,311 dot output position control unit 212 dither threshold value selection unit 313 intra-area pixel addition unit 1011 template ROM
1021 Correction value ROM
2121 Dither threshold ROM
3111 Density pattern ROM

Claims (18)

An image processing apparatus that determines a dot-on pixel ratio according to a density value of input multi-valued image data, and outputs as a low-value image data having a smaller number of multi-values than the multi-valued image data,
Shape recognition means for recognizing the shape by pattern recognition for the input multi-value image data;
Dot output order determination means for determining the dot output order based on the shape recognized by the shape recognition means and the image data around the target pixel;
Quantization means for quantizing the input multi-value image data into low-value image data;
Dot output position control means for controlling the dot-on pixel position according to the density value of the input multi-value image data,
The dot output position control means arranges the dot-on pixels based on the dot output order determined by the dot output order determination means. Edge level calculating means for calculating an edge level from the input multi-valued image data;
The dot output position control means arranges the dot-on pixels based on the dot output order determined by the dot output order determination means and the edge level calculated by the edge level calculation means. Item 6. The image processing apparatus according to Item 1. The shape recognition means recognizes the shape of a line drawing of the input multi-value image data,
The image processing apparatus according to claim 1, wherein the dot output order determination unit determines a dot output order based on the shape of the line drawing recognized by the shape recognition unit. An image processing apparatus for outputting input low-resolution multi-value image data as high-resolution low-value image data having a multi-value number smaller than that of the multi-value image data,
Shape recognition means for recognizing the shape by pattern recognition for the input multi-value image data;
Dot output order determination means for determining the dot output order based on the shape recognized by the shape recognition means and the image data around the target pixel;
Quantization means for quantizing the input multi-value image data into low-value image data in units of low resolution;
Dot output position control means for converting the low-resolution low-value image data quantized by the quantization means into the number of dot-on pixels of the high-resolution low-value image data and controlling the dot-on pixel position. And
The dot output position control means arranges dot-on pixels of the high-resolution low-value image data based on the dot output order determined by the dot output order determination means. Edge level calculating means for calculating an edge level from the input multi-valued image data;
The dot output position control means arranges the dot-on pixels based on the dot output order determined by the dot output order determination means and the edge level calculated by the edge level calculation means. Item 5. The image processing apparatus according to Item 4. The shape recognition means recognizes the shape of the line drawing of the input multi-value image data,
The image processing apparatus according to claim 4, wherein the dot output order determination unit determines a dot output order based on the shape of the line drawing recognized by the shape recognition unit. An image processing apparatus for outputting input low-resolution multi-value image data as high-resolution low-value image data having a multi-value number smaller than that of the multi-value image data,
Shape recognition means for recognizing the shape by pattern recognition for the input multi-value image data;
Dot output order determination means for determining the dot output order based on the shape recognized by the shape recognition means and the image data around the target pixel;
Quantization means for quantizing the input multi-value image data into low-value image data by multi-value error diffusion in units of low resolution pixels;
Dot output position control means for converting the low-resolution low-value image data quantized by the quantization means into the number of dot-on pixels of the high-resolution low-value image data and controlling the dot-on pixel position. And
The dot output position control means arranges dot-on pixels of the high-resolution low-value image data based on the dot output order determined by the dot output order determination means. Edge level calculating means for calculating an edge level from the input multi-valued image data;
The quantization unit uses the dither threshold value matrix selected according to the edge level calculated by the edge level calculation unit, and multi-level error diffusion of the input multi-level image data in units of low resolution pixels. The image processing apparatus according to claim 7, wherein the image processing apparatus quantizes the low-value image data.   The dot output position control means arranges the dot-on pixels based on the dot output order determined by the dot output order determination means and the edge level calculated by the edge level calculation means. Item 9. The image processing apparatus according to Item 8. The shape recognition means recognizes the shape of the line drawing of the input multi-value image data,
The image processing apparatus according to claim 8, wherein the dot output order determination unit determines a dot output order based on the shape of the line drawing recognized by the shape recognition unit.   The image processing apparatus according to claim 8, wherein the dither threshold value matrix includes at least one fixed threshold value matrix that takes the same value in all regions in the matrix. The dot output position control means, when the edge level calculated by the edge level calculation means is larger than a prescribed value, based on the dot output order determined by the dot output order determination means, the high-resolution low-value image Place dot-on pixels of data,
When the edge level calculated by the edge level calculation means is smaller than a prescribed value, the dot-on pixels of the high-resolution low-value image data are not arranged based on the dot output order determined by the dot output order determination means The image processing apparatus according to claim 2, wherein the image processing apparatus is an image processing apparatus.   The dot output position control means arranges the dot-on pixels of the high-resolution low-value image data based on another dither threshold matrix when the edge level calculated by the edge level calculation means is smaller than a prescribed value. The image processing apparatus according to claim 12. The shape recognition means performs pattern recognition using template matching for comparing a template stored in advance in memory with a part of the multi-valued image data,
The image processing apparatus according to claim 1, wherein the dot output order determination unit determines a dot output order based on a matched template and image data around a target pixel.   Using the pattern recognition result by the shape recognition means, the input multivalued image data is subjected to jaggy reduction processing, and the multilevel image data after the jaggy reduction processing is subjected to edge level calculation processing and low value image data. The image processing apparatus according to claim 1, wherein a quantization process is performed on the image processing apparatus. An image processing method for determining a dot-on-pixel ratio according to a density value of input multi-valued image data, and outputting with a small value image data having a smaller number of multi-values than the multi-valued image data,
Recognize the shape by pattern recognition for the input multi-value image data,
Based on the recognized shape and the image data around the pixel of interest, determine the dot output order,
Quantize input multi-value image data into low-value image data,
Control the dot-on pixel position according to the density value of the input multi-value image data,
An image processing method comprising arranging the dot-on pixels based on a determined dot output order. An image processing method for outputting input low-resolution multi-value image data as high-resolution low-value image data having a multi-value number smaller than that of the multi-value image data,
Recognize the shape by pattern recognition for the input multi-value image data,
Based on the recognized shape and the image data around the pixel of interest, determine the dot output order,
Quantize the input multi-value image data into low-value image data in units of low resolution pixels,
The quantized low-resolution low-value image data is converted into the number of dot-on pixels of the high-resolution low-value image data to control the dot-on pixel position,
An image processing method, comprising: arranging dot-on pixels of the high-resolution low-value image data based on the determined dot output order. An image processing method for outputting input low-resolution multi-value image data as high-resolution low-value image data having a multi-value number smaller than that of the multi-value image data,
Recognize the shape by pattern recognition for the input multi-value image data,
Based on the recognized shape and the image data around the pixel of interest, determine the dot output order,
Quantize input multi-value image data into low-value image data by multi-value error diffusion in low resolution pixel units,
The quantized low-resolution low-value image data is converted into the number of dot-on pixels of the high-resolution low-value image data to control the dot-on pixel position,
An image processing method, comprising: arranging dot-on pixels of the high-resolution low-value image data based on the determined dot output order.

Images