KR100484170B1 - Method and apparatus for improvement of digital image quality - Google Patents

Abstract

A method for improving the image quality of a document image composed of pixels having a predetermined resolution obtained by scanning an original mixed with a background, a character, and a picture is disclosed. The method comprises the steps of: (a) converting the color data of the pixel of interest to a predetermined color space to obtain lightness and saturation components; and (b) low pass filtering of the pixel of interest according to the size of the bimodality value of the pixel of interest. (C) history information according to the pixel type of the pixel of interest determined from the lightness and saturation components obtained in step (a) and the connection components around the pixel of interest preprocessed in step (b). (B) detecting the block feature type of the pixel of interest using the pixel type of the pixel of interest determined in step (c) and the history information of the pixel of interest updated (e) The region type to which the adjacent pixel on the left side of the pixel belongs, the region type of the pixel adjacent to the pixel of interest in the upper line of the pixel of interest, the pixel type of the pixel of interest determined in step (c), and (d) The composed of a note determining a region type belonging to a target pixel according to the block type of the pixel characteristic, and (f) (e) performing a different image quality improvement process with each other in accordance with the determined region types in a step detected by the.

Description

Method and apparatus for improving digital image quality {Method and apparatus for improvement of digital image quality}

The present invention relates to a method and apparatus for improving digital image quality. In particular, when copying a document mixed with text and pictures, the scanned document image is divided into a text area, a background area, a halftone picture area, and a continuous tone picture area. The present invention relates to a digital image quality improving method and apparatus for improving the overall image quality of an output by applying an image quality improvement process suitable for each area type.

In general, the interest in the method of effectively representing the original document using only black and white binary information has been increased with the development of copiers, facsimile machines or multi-function devices. Dithering method is used to express actual continuous brightness as pseudo brightness, and Bi-level segmentation which properly divides text and background for easy reading of text for text documents. .

However, these methods may not be an efficient method for a document having a mixture of characters and images, which are commonly encountered in practice. That is, in the case of applying the binary division method, a problem such as a false contour occurs in the image. In methods, pseudo-brightness tends to smooth the edges and degrade the readability of character parts. In order to solve this problem, various researches are being conducted to improve the quality of mixed documents.

1 is a flowchart illustrating an example of a conventional method for improving image quality of mixed documents, and is a U.S.P. document issued to Ramin Khorram. 6,175,425 (Document imaging system for autodiscrimination of text and images, Jan. 16, 2001).

Referring to FIG. 1, a method of improving mixed document image quality includes receiving a multi-valued signal representing a document image input from a scanner or another input device (100), and generating a template defining a high contrast region in a document. Applying a filter (102), separating the character and the picture area (104), applying a high pass filter to the text area to improve the image of the text area (106), applying a low pass filter to the picture area A process of improving an image region image (108), a process of generating an output image by connecting the processed character region image and the image region image (110), a process of halftone processing the output image (112), and a half A process 114 of outputting the tone image is performed.

In the above method, the input document image is divided into blocks, and then a high pass filter is applied to each block, thereby dividing the input document image into a character area having many high band components and a picture region having few high band components. Therefore, when two adjacent blocks are different from each other, a blocking phenomenon may occur, and in order for all pixels included in the block to be the same area, the block size must be very small. Extraction becomes difficult, which makes it difficult to determine the appropriate block size.

Meanwhile, U.S.P. issued to Yee Seung Ng. 6,078,697 (Method and apparatus for segmenting image data into contone, text and halftone classifications, Jun. 20, 2000) sets a window of a constant size around each pixel of each input document image A gradient operation is applied to the pixels present. The fuzzy probabilities, which are values representing the likelihood that the center pixel belongs to the character region, the halftone region, and the continuous tone region, are calculated by using the size and direction of the resulting gradient. The class of the center pixel is determined by comparing three fuzzy probabilities and determining the class of the center pixel as a region having the largest value, and finally, the class of the center pixel is determined through a post-processing process for re-dividing the misidentified pixels. In this method, many errors can be generated because the class of the center pixel is determined using only the non-wide information. Also, the neighboring pixels are judged to be different classes so that different improvement methods can be applied to neighboring pixels. This will result in unobtrusive results.

In addition, U.S.P. issued to Hakan Ancin. 5,956,468 (Document segmentation system, Sep. 21, 1999) converts the input document image into a low resolution image to find large characters and picture areas in the low resolution image, and dark brightness characters on a bright background in other areas. It finds the area and performs the quality improvement process to improve the readability of the text area. This method has the disadvantage of emphasizing only dark text areas written on a light background and hardly emphasizing the picture areas.

The technical problem to be solved by the present invention is to copy a document having mixed text and pictures, and divide the scanned document image into a text area, a background area, a halftone picture area, and a continuous tone picture area. It is to provide a digital image quality improvement method for improving the overall image quality of the output by applying an appropriate image quality improvement process.

Another object of the present invention is to provide an apparatus most suitable for realizing the digital image quality improvement method.

According to an aspect of the present invention, there is provided a digital image quality improving method comprising: (a) converting a color data into a predetermined color space to obtain a brightness component and a saturation component from color data of a pixel of interest; (b) a pixel preprocessing step of performing low pass filtering on the pixel of interest according to the size of the bimodality value of the pixel of interest; (c) history information according to the pixel type of the pixel of interest determined from the lightness and saturation components obtained in step (a) and the connection components around the pixel of interest preprocessed in step (b); A history information updating step of updating using history information; (d) a block feature type detection step of detecting a block feature type of the pixel of interest using the pixel type of the pixel of interest and the updated history information of the pixel of interest determined in step (c); (e) the region type to which the adjacent pixel in the left direction of the pixel of interest belongs, the region type of the pixel adjacent to the pixel of interest in the upper line of the pixel of interest, the pixel type of the pixel of interest determined in step (c), and the a region type determining step of determining a region type to which the pixel of interest belongs according to the block feature type of the pixel of interest detected in step (d); And (f) performing different image quality improving processes in accordance with the area type determined in step (e).

In order to achieve the above technical problem, the digital image quality improving apparatus according to the present invention receives color data of pixels converted into a predetermined color space for each line and obtains brightness and saturation components of the pixel of interest from the color data of the pixel of interest. A pixel input unit for each line; Low-pass filtering is performed on the pixel of interest according to the size of the bimodality value of the pixel of interest provided from the pixel input unit for each line, and from the lightness and saturation components of the pixel of interest and the connection component around the pixel of interest. A history information updating unit for updating history information according to the determined pixel type of the pixel of interest using history information before the pixel of interest; A background block feature detection unit for dividing the pixel of interest into pixels having a background block feature when a block having a first size includes a background pixel in a left direction and an upward direction of the pixel of interest with reference to the history information; A picture block feature detection unit for dividing the pixel of interest into pixels having a picture block feature when a block having a second size is formed of a picture pixel in a left direction and an upward direction of the pixel of interest with reference to the history information; It consists of a line memory having a width of 1 bit, and when the picture block feature is detected in the pixel of interest by the picture block feature detection unit, the next position of the position where the background block feature is last detected in the left direction of the pixel of interest in the line memory is detected. A storage unit for setting and storing the data as '1'; The pixel of interest is divided into pixels having a halftone block by using a pixel type of the pixel of interest, a pixel type of pixels existing in a left direction of the pixel of interest, and an area type of an adjacent pixel in an upper direction of the pixel of interest. A halftone block feature detector; A region type to which the adjacent pixel in the left direction of the pixel of interest belongs, a pixel type of the pixel of interest provided from the history information updater, and an area type of a pixel adjacent to the pixel of interest in an upper line of the pixel of interest, and each block An area type determination unit that determines an area type to which the pixel of interest belongs according to the block feature type of the pixel of interest provided from the feature detection unit and information of the line memory of the storage unit; And an image quality improvement processing unit which performs image quality improvement processing by varying the degree of improvement according to the area type determined by the area type determination unit.

Next, a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings.

2 is a flowchart illustrating a method for improving digital image quality according to an embodiment of the present invention. The color data conversion step 210 of the pixel of interest 210, history information update step 220 of the pixel type of the pixel of interest, and A block feature type determination step 230, a region type determination step 240 of a pixel of interest, and an image quality improvement processing step 250 for each region type are performed.

In the color data conversion step 210, RGB color data of the pixel of interest is converted into color data having a brightness component, a saturation component, and a color component. At this time, the color model used may include YIQ or YCbCr.

In the YIQ color model, Y denotes brightness, I and Q denote color components, and when an RGB value is given, the YIQ value may be expressed by Equation 1 below.

Y = 0.29900R + 0.58700G + 0.11400B

I = 0.59600R-0.27500G-0.32100B

Q = 0.21200R-0.52300G + 0.31100B

When RGB color data is converted using the YIQ color model, the brightness component is adopted as the Y component, and the saturation component is the sum of the absolute values of the I and Q components or the root mean square of the I and Q components ( Root Mean Square) is preferably obtained. Color component Can be obtained as

Meanwhile, in the YCbCr color model, Y represents lightness, Cb and Cr represent blue and red degrees, and when an RGB value is given, the YCbCr value may be represented by Equation 2 below.

Y = 0.29900R + 0.58700G + 0.11400B

Cb = -0.16874R-0.33126G + 0.50000B

Cr = 0.50000R-0.41869G-0.08131B

When RGB color data is converted using the YCbCr color model, the brightness component is adopted as the Y component, and the saturation component is the sum of the absolute values of the Cb and Cr components or the root mean square root of the Cb and Cr components ( Root Mean Square) is preferably obtained. Color component Can be obtained as

The history information updating step 220 includes a pixel preprocessing step 221, a pixel type determining step 222, and a first and second history information updating / storing step 223.

In operation 221, the low pass filtering of the pixel of interest is performed according to the size of the bimodality value of the pixel of interest or is processed without output.

In step 222, the brightness and saturation components of each pixel of interest of the current line outputted by the low pass filtering or no processing are compared with the brightness and chroma reference values, respectively. Determine if it is a picture pixel. If the saturation component of the pixel of interest is lower than the saturation reference value S0 and the brightness component is higher than the high brightness reference value B1, the pixel of interest is divided into a background pixel. On the other hand, all pixels that do not satisfy the conditions of the background pixel become pixels instead of the background pixel. If the pixel of interest is a pixel other than the background pixel, the pixel of interest is divided into a picture pixel when the chroma component of the pixel of interest is higher than the chroma reference value S0 or the brightness component is higher than the low brightness reference value B2. Here, the saturation reference value S0 is preferably about 15, and the high brightness reference value B1 is preferably close to white, preferably about 250, and the low brightness reference value B2 is preferably close to black. It is about 100.

On the other hand, in order to determine whether the pixel of interest is a halftone pixel, first, the 4-way connection component values of the pixels existing in the 3 × N window (where N is 3 or more) centered on the pixel of interest, for example, 3 × 3 window It is judged whether it is 2-3, Preferably it is two or more. In addition, when the condition that the four-way connection component value is 2 or more is satisfied, the difference between the maximum value and the minimum value of the brightness components of the pixels existing in the 3 × 3 window centered on the pixel of interest is the brightness difference reference value, preferably 30 or more. Determine the cognition.

In step 223, the first or second history information of the pixel of interest is updated and stored using the determination result of step 222. Here, the first history information has a value indicating how many pixels are continuously background pixels in an upward direction with respect to the center pixel in order to determine the background block feature type, and the second history information is used to determine the picture block feature type. For example, the pixel has a value indicating how many pixels are successively picture pixels in the upward direction with respect to the center pixel.

As a result of the determination in step 222, when the pixel of interest is divided into a background pixel, the number of background pixels consecutive in an upward direction including the pixel of interest is stored as the first history information for the pixel of interest. Preferably, m is stored as first history information of the pixel of interest when the number of background pixels continuous in the upward direction including the pixel of interest is m or more, and m or less is the first history of the pixel of interest. Save as information. Preferably, m may be designated as 3 at 600 dpi (dor per inch) resolution. In this case, the memory used for the background pixel for storing the first history information is [ ], Where [] is a Gaussian symbol, which requires two bits when m is three.

As a result of the determination in step 222, when the pixel of interest is divided into picture pixels, the number of consecutive picture pixels including the pixel of interest in the upward direction is stored as the second history information for the pixel of interest. Preferably, p is stored as second history information of the pixel of interest when the number of consecutive picture pixels, including the pixel of interest, in the upward direction is greater than or equal to p. Save as information. Preferably, p may be designated as 7 at 600 dpi (dor per inch) resolution. In this case, the memory used for the picture pixel for storing the second history information is [ ], 3 is required if p is 7.

Therefore, the total memory capacity used to store the first and second history information for each pixel of one line becomes (number of pixels per line x 5 bits), so that a significant cost reduction in hardware implementation can be obtained.

In the block feature type detection step 230, the background block feature and the picture block feature are detected using the first or second history information. The background block feature is detected with reference to the first history information, in which m is increased to the left of the pixel of interest including the pixel of interest as the background pixel and / or to the right from the pixel adjacent to the line of interest in the previous line as shown in FIG. 3A. Detected when more than n stored background pixels appear in succession. Preferably, m and n may be designated 3 and 5, respectively, at 600 dpi resolution. The picture block feature is detected with reference to the second history information, which includes the pixel of interest including the pixel of interest as shown in FIG. 3B to the left of the pixel of interest including the pixel of interest and / or from a pixel adjacent to the line of interest in the previous line. Detected when more than q picture pixels in which p is stored to the right appear in succession. Preferably, at 600 dpi resolution, p and q may be designated as 7 and 20, respectively. Also, the picture block feature is detected when r or more pixels other than the background appear in the leftward direction of the pixel of interest, including the pixel of interest, as shown in FIG. 3C. Preferably, r is 200 at a 600 dpi resolution. Can be specified.

On the other hand, the halftone block feature is that the pixel of interest is a halftone pixel, and the halftone pixel exists within a halftone distance reference value, preferably 11, in the left direction including the pixel of interest, or a pixel adjacent to the pixel of interest in the upper line If it belongs to the halftone picture region, the halftone block feature is detected.

In the region type determining step 240, the region to which the previous pixel existing in the left direction of the pixel of interest belongs, the region to which the pixel adjacent to the pixel of interest of the previous line belongs, the pixel type in step 222, and the block feature in step 230 Based on the type detection result, the area to which the pixel of interest belongs is determined.

In the image quality improvement process 250, when the region type to which the pixel of interest belongs is determined in step 240, the image quality improvement process is performed by varying the degree of improvement for each region type. An example of the highlighting technique used here is an unsharpen masking process, whereby the brightness of the pixel to be emphasized is obtained by the following equation (3).

Le = L5 + k (L5-Lavg)

Where L5 is the original brightness of the pixel to be emphasized, Lavg is the average value of the brightness within the set window, Le is the highlighted brightness value, and k is the enhancement coefficient.

First, when the pixel of interest is divided into character areas, the brightness of the pixel of interest is divided into three grades of high, middle, and low grades to maximize the difference with the background area. When output, the RGB value is designated as (255, 255, 255) for white filling, and if it belongs to a low grade, it is designated as (0, 0, 0) for black filling. On the other hand, if the brightness of the pixel of interest belongs to the intermediate grade that appears frequently at the boundary between the text area and the background area, non-sharp masking is applied to emphasize the boundary of the text area. In this case, the emphasis coefficient applied may be calculated using average brightness of 3 × 3 neighboring pixels of the pixel of interest, and preferably, (average brightness, emphasis factor) is (100, 6) and (250, 1). Is variable on a straight line connecting the. On the other hand, if the pixel of interest is divided into a background area, the original image is output as it is.

If the pixel of interest is divided into continuous tone picture areas, non-sharp masking is applied to emphasize the boundary of the continuous tone picture areas. In this case, the emphasis coefficient applied can be calculated using the average brightness of 3 × 3 neighboring pixels of the pixel of interest, and preferably, (average brightness, emphasis factor) is (0, 6) and (255, 1). Is variable on a straight line connecting the.

On the other hand, when the pixel of interest is divided into halftone picture regions, the original image is output as it is, or a low pass filter is applied to prevent annoying patterns from appearing, or a very small emphasis coefficient, preferably 1 or less Apply non-sharpening masking processing using.

FIG. 4 is a flowchart illustrating an example of updating the history information 220 in FIG. 2, wherein the pixel selection step 410, the pixel preprocessing step 420, the pixel type determination step 430, and the first history information are shown. An update step 440, a second history information update step 450, and a final pixel determination step 460 are performed.

In the pixel selection step 410, since the history information update process is performed from the first pixel of the current line, in step 411, the initial position value I of the pixel is set to '0'. In operation 412, the position value I of the pixel is increased by '1'.

In the pixel preprocessing step 420, in step 421, the bimodality of pixels existing in a 3 × N window (where N is 3 or more) centered on a pixel of interest, for example, a 3 × 3 window, is determined. Calculate. In step 422, the bimo dollar tee calculated in step 421 is compared with a predetermined bimo dollar tee reference value, and it is determined whether the bimo dollar tee is larger than the reference value. Here, the bimodality reference value can be preferably set to 50.

As a result of the determination in step 422, if the calculated bimodollar is larger than the reference value, the pixel type determination step 430, which is the next step, is performed without performing other processing on the pixel of interest, and the calculated bimodollity is the reference value. In the smaller case, the low pass filtering is performed on the pixel of interest in step 423, and then the pixel type determination step 430 is performed. Here, the bimo dollar tee has a value larger than the reference value means that the pixel of interest is a pixel included in the boundary of the character area or the halftone area, and the bimo dollar tee has a value smaller than the reference value. This means that the pixel is included in the background area or the continuous tone picture area.

In the pixel type determination step 430, in step 431, it is determined whether the saturation component of the I-th pixel is lower than the saturation reference value S0 and the brightness component is higher than the high brightness reference value B1. Determined by pixel. In operation 432, when it is determined that the pixel is not a background pixel, it is determined whether the pixel type of the I-th pixel is a picture pixel. In this case, the saturation component of the I-th pixel is higher than the saturation reference value S0. When the brightness component is larger than the low brightness reference value B2 and smaller than the high brightness reference difference B1, the I-th pixel is divided into picture pixels.

In the background history, that is, the first history information update step 440, in step 441, when the I-th pixel is determined to be the background pixel in step 431, the picture history of the I-th pixel is set to '0'. In step 442, it is determined whether the stored background history is smaller than m, for example, '3'. As a result of the determination in step 442, if the background history is equal to or greater than m, the background history of m is maintained as it is, and if the background history is smaller than m, the background history is increased by '1'.

In the picture history, that is, the second history information update step 450, in step 451, when the I-th pixel is determined to be a pixel that is neither a background pixel nor a picture pixel, in step 432, the background history of the I-th pixel is' Set to 0 '. In step 452, if it is determined in step 432 that the I-th pixel is determined to be a picture pixel, the background history of the I-th pixel is set to '0', and in step 453, the previously stored picture history is p, for example, ' Determine if it is less than 7 '. As a result of the determination in step 453, if the picture history is equal to or larger than p, the picture history of p is maintained as it is, and if the picture history is smaller than p, the picture history is increased by '1'.

In the final pixel determination step 460, it is determined whether the I-th pixel is the last pixel of the current line. If the final pixel is the final pixel, the flow is terminated. If the final pixel is not the final pixel, the flow returns to step 412 to determine the pixel preprocessing step 420. Step 430, the first history information updating step 440, the second history information updating step 450, and the final pixel determination step 460 are repeatedly performed.

Next, a method of calculating bimodality in step 421 of the pixel preprocessing step 420 of FIG. 4 will be described with reference to FIG. 5.

Referring to FIG. 5, the method for calculating bimodality is a window setting step (step 511), an average value calculation step (513 step), a grouping step (515 step), and a bimodality value determination step (517 step) Step 519).

In step 511, a 3x3 window is set around the pixel of interest, and in step 513, an average value of brightness components of all pixels existing in the 3x3 window is calculated.

In step 515, the brightness value of each pixel present in the 3 × 3 window is compared with the average value of the brightness components obtained in step 513, and the first pixel group having the brightness component higher than the average value of the brightness component and the average value of the brightness component are compared. Grouped into a second pixel group having a low brightness component.

In operation 517, an average value of brightness components of pixels belonging to the first pixel group divided in operation 515 and an average value of brightness components of pixels belonging to the second pixel group are calculated. In operation 519, the first and second pixels obtained in operation 517. The difference in the average value of the lightness components of the group is obtained and determined as the bimodality value of the pixel of interest.

Next, a method of obtaining connection component values of pixels existing in a 3x3 window in order to determine halftone pixels in the pixel type determination step 221 of FIG. 2 will be described with reference to FIGS. 6 to 8. do.

Referring to FIG. 6, the method for obtaining a connected component value includes a window setting step (step 611), an average value calculating step (613) for a lightness component, a binarization step (step 615), and a step for determining a connected component value (step 617). .

In step 611, a 3x3 window is set around the pixel of interest, and in step 613, an average value of brightness components of all pixels existing in the 3x3 window is calculated.

In step 615, the brightness value of each pixel existing in the 3 × 3 window is compared with the average value of the brightness components obtained in step 611, and binarization is performed based on a comparison result of '1' or '0'. For example, if the brightness component of the pixel is smaller than the average value of the brightness component, '1' is assigned, and '0' is assigned to the pixel.

In step 617, a logic circuit is used instead of a lookup table to minimize the required memory capacity, and the connection component value CC is determined using a different calculation method according to the binary value of the pixel of interest. This will be described in more detail with reference to FIGS. 7 and 8.

In step 711, it is determined whether the binary value of the pixel of interest, that is, the B5 pixel, is 1.

In step 712, if the binary value of the pixel of interest is '0' as a result of the determination in step 711, it counts how many blocks of '0' and '1' exist in the clockwise direction 80 from the B1 pixel. The value is determined as the connected component value (CC). For example, as illustrated in FIG. 8, when the B1, B3, B4, and B8 pixels are '0' including the B5 pixel of interest, the connection component value CC is 3.

In operation 713, when the binary value of the B5 pixel is '1', the sum of the binary values of the B2, B4, B6, and B8 pixels is defined as a connection degree, and the value is different according to the connection value. The connection component value is determined by the method. First, in step 714, it is determined whether the connectivity is '0', and when the connectivity is '0', the sum of the binary values of the B1, B3, B7, and B9 pixels plus '1' is added to the connection component value ( CC) (step 715). On the other hand, in step 716, if the connection degree is not '0' in step 714, it is determined whether the connection degree is '1', and if the connection degree is '1', the 'continuity in the clockwise direction 60' The number of 1's is counted and the value is determined as the connected component value CC (step 717).

In step 718, when the connection degree is not '1', the determination result in step 716 determines whether the connection degree is '3' or '4', and when the connection degree is '3' or '4', the connected component The value CC is determined as '1' (step 719). On the other hand, in step 720, when the connection degree is '2' in operation 718, it is determined whether the binary values of the B2 and B8 pixels are '1' or the binary values of the B4 and B6 pixels are '1'. If the binary value of the B8 pixel is '1' or the binary value of the B4 and B6 pixels is '1', the connection component value CC is determined as '1' (step 721).

In operation 722, when the binary value of the B2 or B8 pixel is not '1', and the binary value of the B4 or B6 pixel is not '1', it is determined whether the binary value of the B2 pixel is '1'. If the binary value of the B2 pixel is '1', it is determined whether the binary value of the B4 pixel is '1' (step 723), and if the binary value of the B4 pixel is '1', the connection component value CC is set to B9. The binary value of the pixel is determined by adding '1' (step 724). In operation 725, when the binary value of the B4 pixel is not '1' in operation 723, the connection component value CC is determined as the binary value of the B7 pixel plus '1'.

In operation 726, when the binary value of the B2 pixel is not '1', it is determined whether the binary value of the B4 pixel is '1', and when the binary value of the B4 pixel is '1', the connection component value is determined. (CC) is determined as a value obtained by adding '1' to the binary value of the B3 pixel (step 727). In operation 728, when the binary value of the B4 pixel is not '1' in step 726, the connection component value CC is determined as the binary value of the B1 pixel plus '1'.

FIG. 9 is a state diagram illustrating an embodiment of the region type determination step 240 of FIG. 2. Prior to the description, the image mark memory used for determining the continuous tone picture region of the region type determining step 240 will be described.

The image mark memory is a line memory having a width of 1 bit. When the picture block feature is detected and the pixel of interest is determined to be a continuous tone picture area, the image mark memory has the position of the last detected background block feature in the left direction of the pixel of interest in the image mark memory. Set the next position to '1'. This allows the next line to be determined as the continuous tone picture area from the position where the image mark memory is set to '1'.

Referring back to FIG. 9, when the previous pixel belongs to the background area in state 911, if the pixel of interest is a pixel other than the background (913), the pixel of interest is divided into a character area, and if the pixel of interest is a background pixel (912), the pixel of interest Is maintained as the background area. On the other hand, when the previous pixel belongs to the background area in state 911, if the value of the image mark memory is set to '1' (921), the pixel of interest is divided into a continuous tone picture area.

When the previous pixel belongs to the text area in state 914, when the background block feature is detected in the pixel of interest (916), the pixel of interest is divided into the background area. Meanwhile, when the picture block feature is detected in the pixel of interest or the value of the image mark memory is set to '1', the pixel of interest is divided into a continuous tone picture area. Also, in the case where the pixel adjacent to the pixel of interest in the upper line belongs to the continuous tone picture area or the halftone picture area (917), the pixel of interest is divided into the continuous tone picture area. If these conditions are not satisfied (915), the pixel of interest is kept in the character area.

In the state 918, when the previous pixel belongs to the continuous tone picture region, when the background block feature is detected in the pixel of interest (920), the pixel of interest is divided into the background region. In addition, when the halftone block feature is detected in the pixel of interest (922), the pixel of interest is divided into a halftone picture area. If the condition is not satisfied (919), the pixel of interest is maintained in the continuous tone picture area.

In the state 923, when the previous pixel belongs to the halftone picture region, the pixel of interest exists in two conditions for determining the halftone pixel, that is, a condition in which the four-way connection component value is 2 or more, and a 3 × 3 window centered on the pixel of interest. If the difference between the maximum value and the minimum value of the brightness components of the pixels satisfies one of the conditions that must exceed the brightness difference reference value (924), the pixel of interest is determined as the halftone picture area, and if neither of the two conditions is satisfied (925). Determined by continuous tone picture area.

Meanwhile, the above-described embodiments of the present invention can be written as a program that can be executed in a computer, and can be implemented in a general-purpose digital computer that operates the program using a computer-readable recording medium. The computer-readable recording medium includes, for example, magnetic storage media such as ROM, floppy disk, hard disk, etc., optical read media such as CD-ROM, DVD, etc., and storage media such as carrier wave such as transmission over the Internet. .

FIG. 10 is a block diagram showing the structure of an embodiment of a digital image quality improving apparatus according to the present invention, wherein a pixel input unit 1011 for each line, a history information update unit 1012, a background block feature detection unit 1013, and a picture block feature detection unit 1014, a halftone block feature detection unit 1015, an upper line information storage unit 1016, an area type determination unit 1017, an image quality improvement processing unit 1019, and a storage unit 1020.

The pixel input unit 1011 for each line inputs color data of pixels converted into a YIQ or YCrCb color space for each line, and obtains a brightness component, a saturation component, and a color component from the color data of the pixel of interest.

The history information updating unit 1012 updates history information corresponding to the pixel type of the pixel of interest provided from the pixel input unit 1011 for each line using history information before the pixel of interest. To this end, first, the low pass filtering of the pixel of interest or output is performed without processing according to the size of the bimodality value of the pixel of interest, and the brightness of each pixel of interest in the current line that is output through the low pass filtering or no processing The component and the saturation component are compared with the brightness reference value and the saturation reference value, respectively, and the result of the comparison determines whether the pixel of interest is a background pixel or a picture pixel.

The background block feature detection unit 1013 refers to the history information, and n pixels having m as the first reference for determining the background block feature in the left direction of the pixel of interest are the second reference for the background block feature determination. In the case of continuous display, the pixel of interest is divided into pixels having a background block feature. Here, the case where m and n are 3 and 5 respectively is mentioned.

The picture block feature detection unit 1014 refers to the history information, and q is a second reference for determining the picture block feature in which a pixel, which is the first reference for the picture block feature determination in the left direction of the pixel of interest, is stored. In the case of continuous display, the pixel of interest is divided into pixels having a picture block feature. Here, the case where p and q are 7 and 20 respectively is an example.

The halftone block feature detection unit 1015 uses the pixel type of the pixel of interest, the pixel type of pixels existing in the left direction of the pixel of interest, and the region type of the adjacent pixel in the upper direction of the pixel of interest. It is divided into pixels having a block characteristic.

The upper line region information storage unit 1016 stores information about an area type to which a pixel adjacent to the pixel of interest in the upper line of the pixel of interest belongs.

The storage unit 1020 is an image mark memory including a line memory having a width of 1 bit. When the picture block feature is detected in the pixel of interest by the picture block feature detection unit 1014, the background block feature is finally detected to the left of the pixel of interest. The next position after the specified position is set to '1'.

The region type determiner 1017 includes a region type 1018 to which an adjacent pixel on the left side of a pixel of interest recorded in an internal memory (not shown) belongs, and the pixel type of the pixel of interest provided from the history information updater 1012. And the region type of the upper line adjacent pixel provided from the upper line region information storage unit 1016, the block feature type of the pixel of interest provided from the respective block feature detection units 1013, 1014, and 1015, and the storage unit 1020. The area type to which the pixel of interest belongs is determined in accordance with the information of the image mark memory.

The image quality improvement processing unit 1019 performs image quality improvement processing by varying the degree of improvement according to the area type determined by the area type determination unit 1017 to improve the image quality.

As described above, according to the present invention, when copying a document in which text and pictures are mixed, the scanned document image is divided into a text area, a background area, a halftone picture area, and a continuous tone picture area. By applying an appropriate image quality improvement process, the overall image quality of the output can be greatly improved.

In addition, by using the first and second history information to detect the background block feature and the picture block feature, and minimizing the memory capacity required by using the logic circuit to determine the connection component value, significant cost saving effect is realized in hardware implementation. Can be obtained.

Further, in updating the first and second history information, character and halftone picture areas are determined by determining whether or not to perform a preprocessing step such as low pass filtering according to a result of comparing the bimodality value of the pixel of interest with a predetermined reference value. It is possible to increase the accuracy of the classification and to reduce the amount of line memory required by reducing the information used to detect the halftone block feature.

In addition, the image quality is improved by storing the continuous tone picture area information of the previous line by using a line memory having a bit width instead of a backward update, which requires a relatively large amount of memory in the area type determination process and complicates the circuit implementation. Simplify the required hardware without affecting performance.

Although the present invention has been described with reference to the above embodiments, it is merely illustrative, and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. . Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

1 is a flowchart illustrating a conventional method for improving image quality of mixed documents;

2 is a flowchart illustrating an embodiment of a digital image quality improving method according to the present invention;

3A to 3C illustrate a pixel of interest having a picture block and a background block feature type in the block feature type detection step of FIG. 2;

4 is a flowchart for explaining an embodiment of a history information updating step of FIG. 2;

FIG. 5 is a flowchart for explaining a method of calculating bimodality in FIG. 4; FIG.

6 is a flowchart for explaining an embodiment of a method of obtaining a connection component value for halftone pixel determination in the pixel type determination step of FIG. 2;

7 is a flowchart illustrating an embodiment of a method of obtaining a connected component value according to a binary value of a pixel of interest in FIG. 5;

8 is a view for explaining another embodiment of a method of obtaining a connection component value according to a binary value of a pixel of interest in FIG. 5;

9 is a state diagram illustrating an embodiment of a region type determining step in FIG. 2, and

10 is a block diagram showing the configuration of an embodiment of a digital image quality improving apparatus according to the present invention.

* Explanation of symbols for main parts of the drawings

1011 ... pixel input unit for each line 1012 ... history information updater

1013 ... background block feature detection unit 1014 ... picture block feature detection unit

1015 ... halftone block feature detection unit 1016 ... upper line area information storage unit

1017 ... area type determination unit 1019 ... image quality improvement processing unit

1020 ... storage

Claims (24)

A method for improving the image quality of a document image composed of pixels having a predetermined resolution obtained by scanning an original mixed with a background, text, and picture, (a) converting a color component of the pixel of interest into a predetermined color space to obtain a brightness component and a saturation component; (b) updating history information according to the pixel type of the pixel of interest determined from the lightness and saturation components of the pixel of interest and the connection component around the pixel of interest using history information before the pixel of interest; (c) detecting a block feature type of the pixel of interest using the pixel type of the pixel of interest and updated history information of the pixel of interest; (d) an area type to which the adjacent pixel in the left direction of the pixel of interest belongs, an area type of a pixel adjacent to the pixel of interest in an upper line of the pixel of interest, a pixel type of the pixel of interest, and a block feature type of the pixel of interest Thus, determining a region type to which the pixel of interest belongs from among a text area, a background area, a halftone picture area, and a continuous tone picture area; And and (e) performing different image quality improving processes in accordance with the area type of the pixel of interest. The method of claim 1, wherein the method is performed before step (b). (f) further comprising a pixel preprocessing step of performing low pass filtering on the pixel of interest according to the size of the bimodality value of the pixel of interest, wherein step (f) (f1) calculating a bimodality value of the pixel of interest; (f2) comparing the bimodality value with a reference value; And and (f3) performing low pass filtering on the pixel of interest when the bimodality value is smaller than the reference value. The method of claim 2, wherein the step (f1) calculating an average value of brightness components of all pixels existing in a window having a predetermined size centering on the pixel of interest; (f12) grouping the pixels in the window into a first pixel group having a brightness component higher than an average value of the brightness component and a second pixel group having a brightness component lower than an average value of the brightness component; And (f13) determining the difference between the average value of the brightness components of the pixels belonging to the first pixel group and the average value of the brightness components of the pixels belonging to the second pixel group as a bimodality value of the pixel of interest; How to improve. The method of claim 1, wherein step (b) (b1) A pixel for determining whether the pixel of interest is a background pixel or a picture pixel using the brightness component and the saturation component of the pixel of interest, and a pixel for determining whether the pixel of interest is a halftone pixel using the brightness component and the connection component. Type determination step; (b2) if the pixel of interest is determined to be a background pixel, first history information for storing and updating the number of background pixels continuously including the pixel of interest in an upward direction of the pixel of interest as the first history information of the pixel of interest; An update step; And (b3) if the pixel of interest is determined to be a picture pixel, second history information for storing and updating the number of picture pixels continuously including the pixel of interest in an upward direction of the pixel of interest as the second history information of the pixel of interest; Digital image quality improvement method comprising the step of updating. The method of claim 4, wherein step (b1) (b11) dividing the pixel of interest into a background pixel when the brightness component of the pixel of interest is larger than the high brightness reference value and the chroma component is smaller than the chroma reference value; (b12) dividing the pixel of interest into a picture pixel when the brightness component of the pixel of interest is larger than the low brightness reference value and smaller than the high brightness reference value or the chroma component is larger than the chroma reference value; And (b13) obtaining a difference value between a connection component and a brightness component existing in a window having a predetermined size centering on the pixel of interest, and dividing the pixel of interest into halftone pixels by using the same. The method of claim 5, wherein step (b13) (b131) obtaining a connection component of pixels existing in the window; (b132) comparing a difference value between a maximum value and a minimum value of brightness components of the pixels present in the window and a brightness difference reference value; And (b133) the pixel of interest when the connection component obtained in the step (b131) is equal to or greater than a predetermined value and the difference between the maximum value and the minimum value of the brightness component of the pixels is equal to or greater than the brightness difference reference value as a result of the comparison in the step (b132); Determining a halftone pixel as a digital image quality improvement method. The method of claim 6, wherein step (b131) (b1311) performing binarization on all pixels in the window using the brightness component; And (b1312) The result of binarization, wherein the method of obtaining the connected components is obtained by different methods according to the binary values assigned to the pixel of interest. 8. The digital image quality improving method according to claim 7, wherein the step (b1311) obtains an average value of brightness components of all pixels in the window, compares them with brightness components of each pixel, and binarizes the pixels according to the comparison result. 10. The method of claim 8, wherein in step (b1312), when the brightness component of the pixel of interest is greater than the average value, the number of blocks including pixels having a value different from the binary value of the pixel of interest consecutive in the clockwise direction is counted. And determining the count value as a connected component. 10. The method of claim 9, wherein in step (b1312), when the brightness component of the pixel of interest is smaller than the average value, the connection component is determined in different ways according to the magnitude of the binary values of the four-direction pixels of the pixel of interest. How to improve digital image quality. The method of claim 4, wherein in step (b2), m is set as a first criterion for determining the background block feature, and m is less than or equal to m if the number of consecutive background pixels in the upper direction of the pixel of interest is greater than or equal to m. If the number is stored as the first history information. The method of claim 4, wherein in step (b3), p is set as a first criterion for determining a picture block feature, and if p is greater than or equal to p, the p is less than or equal to the number of consecutive drawing pixels in the upper direction of the pixel of interest. If the number is stored as the second history information. The method of claim 1, wherein step (c) (c1) in the case where pixels having m as a first criterion for determining the background block feature appear in the left and right directions of the pixel of interest with reference to the history information, n consecutively appearing as the second criterion for the background block feature determination; Dividing the pixel of interest into pixels having a background block feature; (c2) when the pixel in which p, the first criterion for determining the picture block feature, is stored in the left and right directions of the pixel of interest with reference to the history information, q consecutively appears as the second criterion for the picture block feature determination, Dividing the pixel of interest into pixels having picture block characteristics; And (c3) A pixel having the halftone block feature of the pixel of interest by using the pixel type of the pixel of interest, the pixel type of pixels existing in the left direction of the pixel of interest, and the region type of the adjacent pixel in the upper direction of the pixel of interest; Digital quality improvement method comprising the step of dividing into. The method of claim 13, wherein step (c3) (c31) if the pixel type of the pixel of interest is a halftone pixel, determining whether a halftone pixel exists within a halftone distance reference value in a left direction of the pixel of interest; (c32) if the pixel type of the pixel of interest is a halftone pixel, determining whether an area type of an adjacent pixel in the upper direction of the pixel of interest is a halftone picture region; And and (c33) dividing the pixel of interest into pixels having a halftone block characteristic when one of the conditions of (c31) and (c32) is satisfied. The method of claim 13, wherein step (c) (c4) if the picture block feature is detected in the pixel of interest in step (c2), setting the next position of the position where the background block feature was last detected in the image mark memory to the left in the image mark memory to '1'; Digital quality improvement method further comprising. The method of claim 1, wherein step (d) (d1) if the area type of the adjacent pixel in the left direction is the background area, determining the area type of the pixel of interest as one of a character area and a background area according to the pixel type of the pixel of interest; (d2) When the area type of the adjacent pixel in the left direction is a character area, the area type of the pixel of interest is set as the background area according to the block feature type of the pixel of interest and the area type of the adjacent pixel in the upper direction of the pixel of interest. Determining one of a continuous tone picture area and a text area; (d3) If the area type of the adjacent pixel in the left direction is a continuous tone picture area, the area type of the pixel of interest is one of a background area, a halftone picture area and a continuous tone picture area according to the block feature type of the pixel of interest. Determining; And (d4) when the area type of the adjacent pixel in the left direction is a halftone picture area, determining the area type of the pixel of interest as one of the halftone picture area and the continuous tone picture area according to the pixel type of the pixel of interest; Digital quality improvement method including. 17. The method of claim 16, wherein the step (d1) determines whether the pixel type of the pixel of interest is a background pixel, and if the background pixel is a background pixel, the area type of the pixel of interest is a background area. Digital image quality improvement method to determine the character area. The method of claim 16, wherein step (d2) (d21) if the block feature type of the pixel of interest is a background block feature, determining the pixel of interest as a background area; (d22) determining the pixel of interest as a continuous tone picture region when the block feature type of the pixel of interest is a picture block feature or the value of the image mark memory for a previous line is set to '1'; (d23) if the area type of the adjacent pixel in the upper direction of the pixel of interest belongs to one of the continuous tone picture area and the halftone picture area, determining the pixel of interest as the continuous tone picture area; And (d24) if the condition of step (d21), the condition of step (d22), and the condition of step (d23) do not correspond, determining the area type of the pixel of interest as a character area; How to improve digital image quality. The method of claim 16, wherein step (d3) (d31) if the block feature type of the pixel of interest is a background block feature, determining the pixel of interest as a background area; (d32) if the block feature type of the pixel of interest is a halftone block feature, determining the pixel of interest as a halftone picture region; And and (d33) determining a region type of the pixel of interest as a continuous tone picture region when none of the conditions of the step (d31) and the condition of the step (d32). 17. The method of claim 16, wherein the step (d4) determines whether the pixel type of the pixel of interest is a halftone pixel. If the halftone pixel is the halftone pixel, the area type of the pixel of interest is a halftone picture area. And the area type of the pixel of interest is determined as a continuous tone picture area. The method of claim 1, wherein step (e) (e1) if the area type of the pixel is a character area, applying one of a back process, a black process, and a non-sharpening masking process of the first enhancement coefficient according to the size of the brightness component of the pixel of interest; (e2) if the area type of the pixel is a continuous tone picture area, applying a non-sharpening masking process of a second enhancement factor; (e3) if the region type of the pixel is a halftone picture region, applying one of unprocessed, low pass filtering, and non-sharpened masking of a third emphasis coefficient to the pixel of interest; And and (e4) outputting the pixel of interest as it is when the region type of the pixel is a background region. 22. The method of claim 21, wherein in the step (e1), the first enhancement coefficient is obtained by using average brightness of pixels in a predetermined window including the pixel of interest. 22. The method of claim 21, wherein in the step (e2), the second enhancement coefficient is obtained by using average brightness of pixels in a predetermined window including the pixel of interest. An apparatus for improving the image quality of a document image composed of pixels having a predetermined resolution obtained by scanning an original mixed with a background, text, and picture, A line-by-line pixel input unit configured to receive color data of pixels converted into a predetermined color space for each line and to obtain brightness and saturation components of the pixel of interest from color data of the pixel of interest; A history information updating unit for updating history information according to the pixel type of the pixel of interest determined from the lightness and saturation components of the pixel of interest and the connection component around the pixel of interest using the history information before the pixel of interest; A background block feature detection unit for dividing the pixel of interest into pixels having a background block feature when a block having a first size includes a background pixel in a left direction and an upward direction of the pixel of interest with reference to the history information; A picture block feature detection unit for dividing the pixel of interest into pixels having a picture block feature when a block having a second size is formed of a picture pixel in a left direction and an upward direction of the pixel of interest with reference to the history information; It consists of a line memory having a width of 1 bit, and when the picture block feature is detected in the pixel of interest by the picture block feature detection unit, the next position of the position where the background block feature is last detected in the left direction of the pixel of interest in the line memory is detected. A storage unit for setting and storing the data as '1'; The pixel of interest is divided into pixels having a halftone block by using a pixel type of the pixel of interest, a pixel type of pixels existing in a left direction of the pixel of interest, and an area type of an adjacent pixel in an upper direction of the pixel of interest. A halftone block feature detector; A region type to which the adjacent pixel in the left direction of the pixel of interest belongs, a pixel type of the pixel of interest provided from the history information updater, and an area type of a pixel adjacent to the pixel of interest in an upper line of the pixel of interest, and each block An area type determination unit that determines an area type to which the pixel of interest belongs according to the block feature type of the pixel of interest provided from the feature detection unit and information of the line memory of the storage unit; And And an image quality improvement processing unit for performing an image quality improvement process by varying the degree of improvement according to the area type determined by the area type determination unit.