JP4182982B2 - Image smoothing method and resolution conversion method - Google Patents

Description

  In the present invention, when a predetermined number of narrow lines composed of subdivided pixels forming an output image are generated from each line forming an input image, the color arrangement of lines adjacent to the generation source line is reflected. The present invention relates to an image smoothing method for smoothing an image.

  When the resolution of a black-and-white binary image is increased, if the line that forms the image is simply copied to increase the number of lines, the step at the black-and-white boundary is enlarged as it is and becomes a large step, resulting in a reduction in image quality. . On the other hand, in the following patent document, a 3 × 3 matrix filter is applied to the original image to perform smoothing together.

  However, since the matrix filter only acts on the apex portion of the step portion of the black-and-white boundary, even if it is effective for the boundary having an angle of about 45 degrees as shown in the patent document, it is slightly different from the main scanning direction. There is a problem that the boundary that forms an angle is smoothed only on the stepped portion, and the boundary cannot be reproduced with a smooth straight line.

Already to this problem, the present inventor sequentially selects each of the lines forming the input image, and detects a matrix for detecting the start, end, and continuation of a stage having a height of one line from the selected target line. A step detecting step for extracting a continuous section of the same color that forms a step with the preceding and succeeding lines adjacent to the target line by the pattern matching process used, and a step in which the continuous section of the same color includes different lengths are included. An output image generation step that generates a predetermined number of narrow lines that are partly color-inverted from the target line based on the position and length of the same color continuous section so as to be reconstructed as overlapping narrow lines. Proposed an image smoothing method (Japanese Patent Application No. 2005-262829).
JP-A-8-16775

  The image smoothing method is to smooth all the stages when generating narrow lines from the lines of the input image. However, when the input image includes both a simple binary image such as a character and a pseudo halftone binary image such as a photograph, such smoothing improves the former image quality, The latter problem that the image quality deteriorates remained unsolved. The reason for the deterioration of the image quality in the latter will be briefly explained. In the pseudo halftone binary image, black and white pixels are arranged according to the error diffusion method or the dither method, so that some pixels are changed from black to white. If the color is corrected from white to black, the color-corrected pixel does not follow the error diffusion method or the dither method, and does not correctly reflect the contrast of the original image. In a conventional facsimile machine or the like, it is determined whether the received image is a simple binary image or a pseudo halftone binary image based on the amount of communication data. The function was turned off to prevent image quality deterioration of the pseudo halftone binary image. In this case, the rattling of the characters included in the received image remains as it is.

  Therefore, the present invention has the property that a simple binary image, particularly characters such as kanji, include relatively long line segments, whereas a pseudo halftone binary image such as a photograph includes only relatively short line segments. As a result, an improved image smoothing method is proposed in which smoothing is performed for the former and smoothing is not performed for the latter.

  That is, the image smoothing method according to the present invention reflects the color arrangement of lines adjacent to the generation source line when a predetermined number of narrow lines forming the output image are generated from each line forming the input image. In the image smoothing method for smoothing an image, each line forming an input image is sequentially selected, and the start end, end, and continuation of a stage having a height of one line are detected from the selected target line. An extraction step of extracting a continuous section of the same color that forms a step with the preceding and following lines adjacent to the target line by a pattern matching process using a matrix, and a length calculated from the start and end of the same color continuous section Is longer than a predetermined reference length, it is reconstructed as a predetermined number of steps by overlapping the narrow lines including the steps having different lengths. When the reference length is shorter than the reference length, the narrow line including the same length as that of the narrow line is reconstructed as a single step, and the start line and the end of the same color continuous section are reconfigured from the target line. And a generating step for generating the predetermined number of narrow lines based on the above. The predetermined reference length is preferably about 3 to 8 pixels, but may be changed so that it can be tried a plurality of times.

  According to this image smoothing method, when the same color continuous section is longer than a predetermined reference length, it is reconstructed as a predetermined number of steps by overlapping the narrow lines including the steps having different lengths. When it is shorter than the length, it is reconstructed as a single step having the same length, and as a result, smoothing is performed on the portion of the simple binary image including a relatively long line segment, and the relatively short line Smoothing is not performed on a pseudo-halftone binary image that contains only minutes. For this reason, the image quality of a simple binary image can be improved without degrading the image quality of a pseudo halftone binary image. Furthermore, if the reference length is made variable, by changing the length, it corresponds to the typical length of the line segment included in the simple binary image and pseudo halftone binary image in the input image at that time. It can also be made.

  The image smoothing method according to the present invention will be described below with reference to the drawings.

  An image smoothed by the image smoothing method of the present invention is defined as an aggregate of pixels that form a plurality of lines in the main scanning direction that are continuously arranged in parallel in the sub-scanning direction.

  Here, when defining a subdivision pixel having a size obtained by dividing the pixel into a plurality of sub-scanning directions, and further defining a narrow line as a subdivision pixel row continuous in the main scanning direction, According to the image smoothing method of the present invention, when a predetermined number of narrow lines forming the output image are generated from each of the lines forming the input image, the color arrangement of the lines adjacent to the generation source line is also reflected. Thus, it can be said that the image is smoothed.

  In the image smoothing method of the present invention, each line in the main scanning direction of the input image is set as the attention line Ln, and an area having a predetermined spread is shifted around the attention line Ln as a center line. By sequentially executing a pattern matching process using a comparison matrix that detects the start, end, and continuation of a stage having a length, the line between the target line Ln and at least two pixels of the line adjacent to the target line Ln A detection step for extracting the same color continuous section S forming a step in the step, and when the detected same color continuous section S is longer than a predetermined reference length TL by calculating the length from the start end and the end thereof, the steps having different lengths Are reconstructed as a predetermined number of steps and are shorter than a predetermined reference length TL, the same as the length of the same color continuous section S. A predetermined number of narrow lines LSn are generated from the target line Ln based on the start and end of the same color continuous section S so as to be reconfigured as a single stage by overlapping the narrow lines LSn including the steps. And a generation step to generate. Here, the reference length TL is preferably about 3 to 8 pixels, but may be changed so that it can be tried a plurality of times.

  Further, in the generation step, including a setting step for setting the color ratio of the subdivided pixels to be color-corrected for smoothing, the length calculated from the start and end of the detected same-color continuous section S is a predetermined reference When it is longer than the length TL, the same color continuous section S is reconfigured as a step having a number of steps corresponding to the set color ratio by overlapping narrow lines LSn including steps having different lengths. You may make it produce | generate the narrow line which carried out the color correction of a part of line, and the narrow line which has the same color arrangement as the attention line.

  In this way, for example, when the image is a two-color image of black and white, if black is set to a higher color ratio than white, the same color continuous section of black and white adjacent to each other as the step boundary will be described. On the black side, the number of subdivided pixels, which are reconfigured as a larger number of steps and converted from black to white, is larger than those converted from white to black. Accordingly, it is possible to make smoothing by changing the thickness between the black diagonal line and the white diagonal line, and an image suitable for the purpose of the image and the user's preference can be obtained.

  FIG. 1 is an example of a black and white input image in which the effect of the present invention is remarkably exhibited. A pseudo halftone binary image obtained by binarizing a photographic image by applying an error diffusion method and a simple character 2 And a digitized image. When attention is paid to the tendency of the pixel arrangement in this example, the pseudo-halftone binary image does not have a clear black-and-white boundary. Therefore, the length of the step is about 3 pixels at the longest. On the other hand, the length of a step in a simple binary image is about 10 pixels in a long portion (particularly, pay attention to a portion surrounded by a solid line). That is, the fact that a simple binary image includes relatively long steps, whereas a pseudo halftone binary image includes only relatively short steps, is supported here. In particular, when a simple binary image is a Chinese character, if the character itself contains many horizontal and vertical line segments, and is scanned in a slightly inclined state, a long step (same color continuous section S) from the horizontal line segment is obtained. Naturally it occurs.

Next, the operation of each step constituting the image smoothing method according to the present invention will be described.
FIG. 2 and FIG. 3 are diagrams for explaining the operation of the simple binary image and the pseudo halftone binary image, respectively. Here, it is assumed that two narrow lines are generated from one line of the input image, and the reference length TL for smoothing is 5 pixels.

  In the setting step, a high color ratio of black or white may be set according to the use of the image or the user's preference.

  In the detection step, an area having an area of 5 × 5 pixels with the target line Ln as a center line is shifted along the target line Ln in the main scanning direction, and adjacent to the target line Ln by pattern matching described later. The same color continuous section S that forms a step between the preceding and following lines Ln−1 and Ln + 1 is extracted. At this time, for example, by counting the number of shift movements, it is possible to grasp the start end of the stage as the pixel position on the target line Ln. The information of the same color continuous section S extracted here includes the start position P0, the end position P1, the side (left and right) and the direction (up and down) of the step, the color, and the like.

  The matrix used in the pattern matching is a stage having a step of one line formed by changing the color of a pixel in the middle of a line passing through the center of the matrix, its start end and its periphery, end and its periphery, or In this example, the pattern is a black-and-white pattern corresponding to 25 pixels forming an area of 5 × 5 pixels. With such a matrix, it is possible to extract a continuous portion of pixels that generate at least one line step as a same color continuous section S.

  However, in order to extract the same color continuous section S, it is sufficient that the color can be identified for a part of 25 pixels forming an area contrasted with the matrix. Therefore, each matrix has a contrast pattern for 25 pixels and a detection pattern. You may comprise as a pair with the pattern for a mask for always considering the contrast about a required pixel as a coincidence. It is only necessary to be able to identify some of the pixels forming the area because the matrix that detects the beginning of the black stage in the figure effectively determines the color of the 8 pixels in the area, It can be easily understood from the fact that the detected matrix effectively discriminates the colors of the five pixels in the area.

  In addition, when separately preparing groups of matrices for detecting the start, end, or continuation of a stage for each of black and white, the pixels of the control pattern are color-inverted with each other in the corresponding matrices. It may be related. Alternatively, in the pattern matching, it is possible to make the matrix common to each of black and white by determining whether the colors of the surrounding 24 pixels are the same with respect to the central pixel of the area. is there. Note that the size of the matrix is not limited to 5 × 5 pixels, and can be changed as appropriate.

  Next, the side where the step occurs and the direction will be described. The same color continuous section S includes a case where an upward step is formed on the left side, a case where an upward step is formed on the right side, and a case where a downward step is formed on the left side. In some cases, a downward step is formed on the right side.

  Also, when the width of the diagonal line formed by successive steps in the input image is 1 pixel, the same color continuous section S may form an upward step on the left side and a downward step on the right side. In some cases, an upward step is formed at the same time, and a downward step is formed on the left side.

  In addition, the same color continuous section S may form an upward step on the left side and an upward difference on the right side, or may form a downward step on the left side and a downward step on the right side. . However, in the embodiment, the smoothing process is not performed in these two cases.

  Next, in the generation step, before generating the narrow line LSn # x, based on the position and length of the same-color continuous section S, a color correction section for correcting the color arrangement in the narrow line LSn # x is calculated. Then, the start and end points are registered in the flag memory. That is, the color correction section is calculated in consideration of the set color ratio and the line count x based on the start position P0, end position P1, rising side and direction, and color of the stage, and corresponds to the value of the line count x. The start and end points are registered in the flag memory.

  When the color correction section is registered, the calculated length of the same color continuous section S is compared with the reference length TL. When the same color continuous section S is longer than the reference length TL, the color correction section is registered. On the other hand, when the same color continuous section S is shorter than the reference length TL, the idea of the present invention may be realized without registration.

  Thereafter, the narrow line LSn # x is generated by inheriting the color arrangement of the pixel of the target line Ln as the color arrangement of the subdivided pixel. At this time, the subdivision is performed in the color correction section registered in the flag memory. The image is smoothed by correcting the color of the pixels. The maximum value m of the line count x is not limited, and the input image is increased in resolution in the sub-scanning direction accordingly.

A flag for registering a color correction section is generally as follows.
That is, in order to generate the narrow lines LSn # 0 to LSn # m corresponding to the target line Ln, it corresponds to the case where an upward white step is extracted on the left side or the right side as the same color continuous section S. The flag WU [0] to the flag WU [m] for registering a color correction section for correcting the color of the subdivided pixels of the narrow lines LSn # 0 to LSn # m from white to black and upward on the left side or the right side Corresponding to the case where the black stage is extracted, flags BU [0] to register a color correction section for correcting the color of the subdivided pixels of the narrow lines LSn # 0 to LSn # m from black to white. Corresponding to the case where the flag BU [m] and the downward white step on the left side or the right side are extracted, color correction for correcting the color of the subdivided pixels of the narrow lines LSn # 0 to LSn # m from white to black Flag WL [0] to flag for registering the section Corresponding to the case where WL [m] and a downward black stage on the left or right side are extracted, a color correction section for correcting the color of the subdivided pixels of the narrow lines LSn # 0 to LSn # m from white to black It is sufficient to prepare the flag BL [0] to the flag BL [m] for registering.

  Here, the flag for registering the color correction section (the start and end of the color correction section are registered in the flag memory) and the calculation of the color correction section registered in the flag correction section are concretely described with respect to the example of FIG. Explained. In the example of FIG. 3, since the length of the same color continuous section S is 4 pixels and is shorter than 5 pixels of the reference length TL, it is not necessary to register the color correction section.

  Table 1 shows an example of a flag to be referred to when a black color ratio higher than white is set.

  In this case, when an upward black stage on the left side is extracted as the same color continuous section S, the flag BU [0] indicates the start end of the same color continuous section S and the end that is a half point of the same color continuous section S. Is registered in the flag memory, and when the upward black stage is extracted on the right side instead of the left side, the start point which is a half point of the same color continuous section S and the end point of the same color continuous section S are indicated by the flag BU [0 ] In the flag memory. In addition, when a black step facing downward on the left side is extracted as the same color continuous section S, the flag BL [1] indicates the start end that is a half point of the same color continuous section S and the end of the same color continuous section S. Is registered in the flag memory, and when the downward black stage is extracted not on the left side but on the right side, the flag BL [1 indicates the start end of the same color continuous section S and the end point that is a half point of the same color continuous section S. ] In the flag memory.

  Thereafter, when the narrow line LSn # 0 is generated, if the start / end of the color correction section is registered in the flag BU [0], the subdivision pixels in the section are color-corrected from black to white, When the narrow line LSn # 1 is generated, if the start / end of the color correction section is registered in BL [1], the subdivision pixels in that section are color-corrected from black to white.

  Table 2 shows an example of a flag to be referred to when black has a color ratio lower than that of white.

  In this case, when an upward white step on the left side is extracted as the same color continuous section S, the flag WU [0] indicates the start end of the same color continuous section S and the end that is a half point of the same color continuous section S. Are registered in the flag memory, and when an upward white stage is extracted on the right side instead of the left side, the start point that is a half point of the same color continuous section S and the end point of the same color continuous section S are flag WU [0. ] In the flag memory. Further, when a white step facing downward on the left side is extracted as the same color continuous section S, the start end that is a half point of the same color continuous section S and the end of the same color continuous section S are set as a flag WL [1]. When registered in the flag memory and a downward white stage is extracted on the right side instead of the left side, the flag WL [1] indicates the start end of the same color continuous section S and the end point that is a half point of the same color continuous section S. Is registered in the flag memory.

  Thereafter, when the narrow line LSn # 0 is generated, if the start / end of the color correction section is registered in the flag WU [0], the subdivision pixels in the section are color-corrected from white to black, When the narrow line LSn # 1 is generated, if the start / end of the color correction section is registered in WL [1], the subdivision pixels in that section are color corrected from white to black.

  In this way, the same color continuous section S longer than the reference length TL that is large in the region of the simple binary image shown in FIG. 2 is smoothed, while the same color continuous section shorter than the reference length TL that is often included in the pseudo-halftone binary image. S can be prevented from smoothing. Also, depending on the color ratio setting, either one of the black and white diagonal lines can be made thicker.

  Further, another example of the flag for registering the color correction section and the calculation of the color correction section registered in the flag will be described, that is, an example in which six narrow lines are generated from one line of the input image. Again, when the length of the same color continuous section S is calculated and the length is longer than the reference length TL, the color correction section is calculated and registered in the flag memory, but the length is shorter than the reference length TL. Does not register the color correction section. The setting of the color ratio value rate is the same as the previous example.

  Table 3 shows an example of a flag to be referred to when black has a higher color ratio than white.

  Here, when an upward black stage on the left side is extracted as the same color continuous section S, the flag BU [0] indicates the start end of the same color continuous section S and the end that is a 4/6 point of the same color continuous section S. Is registered in the flag memory, and the start end of the same color continuous section S and the end that is a 3/6 point of the same color continuous section S are registered in the flag memory as the flag BU [1], and the start end of the same color continuous section S, The end that is a 2/6 point of the same color continuous section S is registered in the flag memory as a flag BU [2], and the start end of the same color continuous section S and the end that is a 1/6 point of the same color continuous section S are obtained. Register in the flag memory as flag BU [3]. Further, when an upward black stage is extracted on the right side instead of the left side, the flag memory stores the start end that is a 2/6 point of the same color continuous section S and the end of the same color continuous section S as the flag BU [0]. Are registered in the flag memory as the flag BU [1], and the end of the same color continuous section S is registered at the flag memory at the 4/6 point of the same color continuous section S. A certain start end and the end of the same color continuous section S are registered in the flag memory as a flag BU [2], and the start end which is a 1/6 point of the same color continuous section S and the end of the same color continuous section S are flagged as the flag BU [ 3] is registered in the flag memory.

  When a downward black stage on the left side is extracted as the same-color continuous section S, the flag memory stores the start end of the same-color continuous section S and the end that is a 2/6 point of the same-color continuous section S as a flag BL [2]. Are registered in the flag memory as a flag BL [3], and the start end of the same color continuous section S and the end of the same color continuous section S are registered. The end of S, which is a 4/6 point, is registered in the flag memory as a flag BL [4], and the start end of the same color continuous section S and the end, which is the 5/6 point of the same color continuous section S, are registered in the flag BL [ 5] is registered in the flag memory. Also, when a downward black stage is extracted on the right side instead of the left side, the flag BL [2] is used as the flag memory with the start end that is a 5/6 point of the same color continuous section S and the end of the same color continuous section S as a flag memory. Is registered in the flag memory as a flag BL [3], and the end of the same color continuous section S is registered as a flag BL [3] at the 3/6 point of the same color continuous section S. A certain start end and the end of the same color continuous section S are registered in the flag memory as a flag BL [4], and the start end that is a 2/6 point of the same color continuous section S and the end of the same color continuous section S are flag BL [ 5] is registered in the flag memory.

  When an upward white step on the left side is extracted as the same color continuous section S, the flag memory stores the start end of the same color continuous section S and the end that is 1/6 of the same color continuous section S as the flag WU [0]. Register with. Further, when an upward white step is extracted on the right side instead of the left side, the flag memory stores the start end that is the 5/6 point of the same color continuous section S and the end of the same color continuous section S as the flag WU [0]. Register with.

  When a white step facing downward on the left side is extracted as the same color continuous section S, the flag memory stores the start end of the same color continuous section S and the end that is 1/6 of the same color continuous section S as a flag WL [5]. Register with. Further, when a downward white step is extracted on the right side instead of the left side, the flag memory stores the start end which is the 5/6 point of the same color continuous section S and the end of the same color continuous section S as the flag WL [5]. Register with.

  Thereafter, the narrow line LSn # x is generated with reference to the corresponding flag, but the same color continuous section S from which the black stage is extracted is generated by the generated narrow lines LSn # 0 to LSn # 4, or By overlapping LSn # 1 to LSn # 5, it is reconfigured as five steps.

  Table 3 shows an example of a flag to be referred to when the standard color ratio is set.

  Here, when an upward black stage on the left side is extracted as the same color continuous section S, the flag BU [0] indicates the start end of the same color continuous section S and the end that is a 2/6 point of the same color continuous section S. Are registered in the flag memory, and the start end of the same color continuous section S and the end of the same color continuous section S are registered in the flag memory as a flag BU [1]. Further, when an upward black stage is extracted on the right side instead of the left side, the flag memory stores the start end that is the 4/6 point of the same color continuous section S and the end of the same color continuous section S as the flag BU [0]. And the start end, which is a 5/6 point of the same color continuous section S, and the end of the same color continuous section S are registered in the flag memory as a flag BU [1].

  When a downward black stage on the left side is extracted as the same color continuous section S, the flag BL [4] is used as the flag memory with the start end of the same color continuous section S and the end of the same color continuous section S as the 1/6 point. And the start end of the same color continuous section S and the end of 2/6 of the same color continuous section S are registered in the flag memory as a flag BL [5]. Also, when a downward black step is extracted on the right side instead of the left side, the flag BL [4] is used as the flag memory with the start end, which is a 5/6 point of the same color continuous section S, and the end of the same color continuous section S as a flag memory. And the start end, which is a 4/6 point of the same color continuous section S, and the end of the same color continuous section S are registered in the flag memory as a flag BL [5].

  When an upward white step on the left side is extracted as the same color continuous section S, the flag memory stores the start end of the same color continuous section S and the end that is a 3/6 point of the same color continuous section S as the flag WU [0]. Are registered in the flag memory as a flag WU [1], and the start end of the same color continuous section S and the end of the same color continuous section S are registered. The end point which is 1/6 of S is registered in the flag memory as flag WU [2]. Further, when an upward white step is extracted on the right side instead of the left side, the flag memory stores the start end, which is a 3/6 point of the same color continuous section S, and the end of the same color continuous section S as a flag WU [0]. Is registered in the flag memory as a flag WU [1], and the end of the same color continuous section S is registered as a flag WU [1] at the 5/6 point of the same color continuous section S. A certain start end and the end of the same color continuous section S are registered in the flag memory as a flag WU [2].

  When a white step facing downward on the left side is extracted as the same color continuous section S, the flag memory uses the start end of the same color continuous section S and the end that is 1/6 of the same color continuous section S as a flag WL [3]. Are registered in the flag memory as a flag WL [4], and the start end of the same color continuous section S and the end of the same color continuous section S are registered. The end point of 3/6 of S is registered in the flag memory as flag WL [5]. Further, when a downward white step is extracted on the right side instead of the left side, the flag memory stores the start end which is the 5/6 point of the same color continuous section S and the end of the same color continuous section S as the flag WL [3]. Is registered in the flag memory as flag WL [4], and the end of the same color continuous section S is registered as a flag WL [4] at the 3/6 point of the same color continuous section S. A certain start end and the end of the same color continuous section S are registered in the flag memory as a flag WL [5].

  Thereafter, the narrow line LSn # x is generated with reference to the corresponding flag, and the same color continuous section S from which the black stage is extracted is generated by the generated narrow lines LSn # 0 to LSn # 2, or By overlapping LSn # 3 to LSn # 5, it is reconfigured as a three-step staircase.

  Table 5 shows an example of a flag to be referred to when the standard color ratio is set.

  Here, when an upward black stage on the left side is extracted as the same color continuous section S, the flag BU [0] indicates the start end of the same color continuous section S and the end that is a 1/6 point of the same color continuous section S. Is registered in the flag memory. Further, when an upward black stage is extracted on the right side instead of the left side, the flag memory stores the start end which is the 5/6 point of the same color continuous section S and the end of the same color continuous section S as the flag BU [0]. Register with.

  When a downward black stage on the left side is extracted as the same-color continuous section S, the flag memory uses the start end of the same-color continuous section S and the end that is 1/6 of the same-color continuous section S as a flag BL [5]. Register with. Also, when a downward black stage is extracted on the right side instead of the left side, the flag BL [5] is set as the flag BL [5] at the start end that is the 5/6 point of the same color continuous section S and the end of the same color continuous section S. Register with.

  When an upward white step on the left side is extracted as the same color continuous section S, the flag memory stores the start end of the same color continuous section S and the end that is a 4/6 point of the same color continuous section S as the flag WU [0]. Are registered in the flag memory as a flag WU [1], and the start end of the same color continuous section S and the end of the same color continuous section S are registered. The end of 2/6 of S is registered in the flag memory as a flag WU [2], and the start of the same color continuous section S and the end of 1/6 of the same color continuous section S are flag WU [ 3] is registered in the flag memory. Further, when an upward white stage is extracted on the right side instead of the left side, the flag memory stores the start end that is a 2/6 point of the same color continuous section S and the end of the same color continuous section S as a flag WU [0]. Is registered in the flag memory as a flag WU [1], and the end of the same color continuous section S is registered as a flag WU [1] at the point of 4/6 of the same color continuous section S. A certain start end and the end of the same color continuous section S are registered in the flag memory as a flag WU [2], and the start end that is a 1/6 point of the same color continuous section S and the end of the same color continuous section S are flag WU [ 3] is registered in the flag memory.

  When a white step facing downward on the left side is extracted as the same color continuous section S, the flag memory uses the start end of the same color continuous section S and the end that is a 2/6 point of the same color continuous section S as a flag WL [2]. Are registered in the flag memory as a flag WL [3], and the start end of the same color continuous section S and the end of the same color continuous section S are registered. The end of S, which is a 4/6 point, is registered in the flag memory as a flag WL [4], and the start end of the same color continuous section S and the end, which is a 5/6 point of the same color continuous section S, are flag WL [ 5] is registered in the flag memory. Also, when a downward black stage is extracted on the right side instead of the left side, the flag memory stores the start end which is the 5/6 point of the same color continuous section S and the end of the same color continuous section S as the flag WL [2]. Is registered in the flag memory as flag WL [3], and the end of the same color continuous section S is registered as a flag WL [3] at the point of 3/6 of the same color continuous section S. A certain start end and the end of the same color continuous section S are registered in the flag memory as a flag WL [4], and the start end that is a 2/6 point of the same color continuous section S and the end of the same color continuous section S are registered in the flag WL [ 5] is registered in the flag memory.

  After that, the narrow line LSn # x is generated with reference to the corresponding flag, but the same color continuous section S from which the black stage is extracted is generated by the generated narrow line LSn # 0, LSn # 1, or By overlapping LSn # 4 and LSn # 5, it is reconfigured as a two-step staircase.

  It should be noted that the specific examples described above are merely examples described, and the calculation method of the color correction section can be freely changed, and should be determined in consideration of the appearance of the output image.

  FIG. 4 shows that the same color continuous section S from which black stages are extracted is divided into four stages of narrow lines LSn # 0 to LSn # 3 or LSn # 2 to LSn # 5 by six narrow lines. It is a figure explaining the method to reconfigure | reconstruct, and has shown the flag referred when producing | generating each narrow line LSn # 0-LSn # 5.

Next, an example of an apparatus that executes the above-described image smoothing method of the present invention will be described.
FIG. 5 is a block diagram illustrating a schematic configuration of an example of an apparatus.

  This image smoothing method apparatus M extracts the same color continuous section S that forms a step of one line height from the input image, and the position of the same color continuous section S detected by the stage detection means 11. A color correction section is calculated based on the length, and the output image generation means 12 generates narrow lines LSn # 0 to LSn # m obtained by correcting the color correction section from the target line Ln, and the color for smoothing. The color ratio register 13 sets the color ratio of the subdivided pixels to be corrected, and the reference length register 14 registers the length of the line segment to be smoothed.

  The stage detection unit 11 is extracted by a line memory unit 11a configured by a shift register or the like that receives and shifts pixel data of an input image for each control clock, and an area extraction unit 11b set in the line memory unit 11a. A pattern matching unit that performs pattern matching on the pixel data in the area and extracts the same color continuous section S.

  The output image generation unit 12 calculates a color correction section based on the start and end of the same color continuous section S detected by the stage detection unit 11, and the color correction section calculation unit 12a. A line counter unit 12b that outputs a line count x that specifies one of the narrow lines LSn # 0 to LSn # m, a flag memory BU (12c) that registers the calculated color correction section, and a flag memory WU (12d) ), A flag memory BL (12e), a flag memory WL (12f), logic gates G1 and G2, and a color correction unit 12g that receives pixels of the input image and outputs segmented pixels of the output image.

The example of the input image of black and white two colors which the effect of this invention is exhibited notably. The figure explaining the effect | action of the step which comprises the image smoothing method of this invention. Another drawing explaining the effect | action of the step which comprises the image smoothing method of this invention. The figure explaining the method to reconfigure | reconstruct the same color continuous area. The block diagram explaining the schematic structure of the apparatus to which this invention is applied.

Explanation of symbols

14 Reference length register Ln Attention line LSn Narrow line LT Reference length M Image smoothing device S Same color continuous section

Claims (2)

Image smoothing that smoothes the image by reflecting the color arrangement of the lines adjacent to the source line when a predetermined number of narrow lines that form the output image are generated from each line that forms the input image In the method
Each line forming the input image is sequentially selected and adjacent to the target line by a pattern matching process using a matrix that detects the start, end, and continuation of a stage having a height of one line from the selected target line. An extraction step of extracting continuous sections of the same color that form a step with the previous and subsequent lines;
When the same color continuous section is longer than a predetermined reference length by calculating the length from its start and end, it is reconfigured as a predetermined number of steps by overlapping the narrow lines including the steps having different lengths, When the length is shorter than the predetermined reference length, from the line of interest, the beginning of the continuous section of the same color is reconfigured as a single stage by overlapping the narrow lines including the stage equal to the length. And a generating step for generating the predetermined number of narrow lines based on the end. In claim 1,
An image smoothing method, wherein the predetermined reference length is changeable.

Images