JPH07162857A - Picture quality adjusting method - Google Patents

Abstract

PURPOSE:To improve picture quality by eliminating the distortion of an image generated due to the compression of a digitized image signal at every picture element block. CONSTITUTION:The image signal is stored in step S210. The fact that a processing target picture element is the picture element of the corner of the picture element block is identified based on a stored image signal in step S220. When the processing target picture element is the picture element of the corner of the picture element block, it is detected whether or not the picture element is the one with prominent distortion in step S230. The picture element value of the picture element is smoothed for the picture element value of an adjacent picture element only when the processing target picture element is the picture element with prominent distortion in step S240.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention eliminates image distortion caused by compressing a digitized image signal,
The present invention relates to an image quality adjustment method for improving image quality.

[0002]

2. Description of the Related Art Conventionally, as a technique in such a field,
For example, some documents were described in the following documents. Reference 1: IEICE Fall National Conference Paper D-60,
(1989) Yasuhiro Takishima Masahiro Wada “Study on Post Filter for Low Bit Rate Video Coding” FIG. 2 is a flowchart showing a conventional image quality adjustment method as shown in Reference 1, which is a digitized original image. Signal is compressed for each pixel block composed of N × M (N and M are positive integers) pixels using discrete cosine transform (hereinafter referred to as DCT transform), and inverse discrete cosine transform (hereinafter referred to as IDCT) is performed. The image quality adjustment method for removing the distortion caused by the compression from the image signal formed by decompression using the conversion) and generating the image signal for reproduction is shown. In the image quality adjustment method of FIG. 2, a step S110 of storing an image signal for each pixel in a storage device and a step of determining whether or not the input image signal of a pixel is an image signal of a pixel near a block boundary of a pixel block. S120 and step S130 of detecting whether the image signal of the pixel near the block boundary is the contour (edge) of the image are sequentially performed. In the image quality adjustment method of FIG. 2, subsequently, a step S140 of filtering an image signal of a pixel near a block boundary other than an edge, and the series of steps S11.
It is composed of step S150 of changing the pixel to be processed from 0 to S140 to the next pixel, and step S160 of judging whether or not the image signal has ended.

Next, a schematic flow of the image quality adjusting method of FIG. 2 will be described. In the image quality adjustment method of FIG. 2, the following processes (1) to (6) are performed in order to remove the distortion of the reproduced image and improve the image quality. (1) Image signal storage processing In step S110, the image signal for each pixel generated through the DCT conversion and the IDCT conversion is sequentially stored in, for example, a storage device. (2) Pixel Discrimination Processing In step S120, it is discriminated whether the image signal of the pixel read from the storage device is a pixel at the boundary of the pixel block, for example, based on the storage address of the storage device. As a result, if the image signal is not the pixel at the boundary of the pixel block, the process proceeds to processing of the next pixel in step S150. If it is the pixel at the boundary of the pixel block, the process proceeds to step S130. (3) Edge Detection Processing In step S130, it is detected whether or not the image signal of the pixel identified as the pixel on the boundary of the pixel block corresponds to the edge. When it is determined that the pixel corresponds to the edge, step S150 is performed, the process returns to step S120 of the pixel identification process again, and the process proceeds to the next pixel. On the other hand, if the target pixel is not an edge, the process proceeds to the filtering process of step S140. (4) Filtering process Only the pixel values of pixels that are on the block boundary and that are not edges are subjected to the filtering process of step S140 and smoothed. After the processing of step S140 ends, the processing proceeds to step S150. (5) Pixel change processing After the filter processing of step S140 is completed, step S140
150 is performed, the process returns to step S120 of the pixel identification process again, and the process proceeds to the next pixel. (6) Image signal end determination processing In step S160, it is determined whether the input of the image signal is completed. If it is determined that it has not ended, the process returns to step S120, and if it is determined that it has ended, a series of processes ends in step S170.

FIG. 3 is a diagram showing pixels and filter processing positions used in the conventional edge detection processing, and FIG. 4 is a flowchart showing the conventional edge detection processing. In FIG. 3, a part of the image to be processed is magnified. In FIG. 3, circles drawn by solid lines or broken lines represent pixels of the image to be processed, and solid lines drawn vertically indicate boundaries of pixel blocks. Further, solid line circle marks B and C represent pixels in the block boundary portion, and broken line circle marks A and D are used together with pixels B and C in the block boundary portion to detect whether the block boundary is an edge. Represents the pixels that can be displayed. Next, the details of the above-mentioned edge detection processing (3) will be described with reference to FIGS. 3 and 4. (3-1) Extraction process For example, when the pixel B to be processed is identified as a pixel on the boundary of the block as a result of step S120 of the pixel identification process, the positional relationship of FIG. 3 is determined in step S131 of FIG. The pixels A, C, and D that it has are extracted, and at the same time, the pixel values a, b, c, and d of those four pixels A, B, C, and D that sandwich the block boundary are obtained. (3-2) Difference calculation process The calculation process of step S132 is performed, and each pixel value a,
The absolute difference values α1 to α3 are obtained from b, c, and d by calculating the following equation.

[0005] Here, for example, | bc | represents the absolute value of the difference between the pixel value b and the pixel value c. (3-3) Comparison processing In order to detect whether the pixel B is an edge, step S
The absolute difference values α1 to α3 obtained by the process of 132 are
Each of the thresholds TH1 set in steps S133 to S135
Or compared to TH2. That is, in step S133, the absolute difference value α1 is compared with the threshold value TH1,
At 134 and 135, each absolute difference value α2 and α3 is compared to a threshold TH2. However, the threshold value TH1 <TH2. When α1 <TH1 in step S133, the process proceeds to step S134. In step S134, the absolute difference value α2 and the threshold value TH2 are compared, and α2 ≦
If TH2, the process proceeds to step S135. In step S135, the absolute difference value α3 and the threshold value TH2 are compared, and when α3 ≦ TH2, it is determined that the pixel B is not an edge. With this determination, the process proceeds to the filtering process of step S140, and the subsequent processes are performed. If even one of the comparison processes in steps S133 to 135 is not satisfied, it is determined that pixel B is an edge, and the process proceeds to step S250. Although FIG. 3 shows the case of the block boundary in the vertical direction, the same processing is performed also in the case of the block boundary in the horizontal direction to detect whether the pixel B is an edge.

FIG. 5 is a flowchart showing a conventional filter process, and the process of step S240 will be described with reference to FIGS. 3 and 5. In the extraction processing of step S131, the pixel values b of the pixels B and C that sandwich the extracted boundary,
c is read in step S141, and step S142
Then, those pixel values b and c are smoothed by filtering. By performing the above processing, the distortion of the reproduced image generated by compressing the digitized image signal is removed.

[0007]

However, the conventional image quality adjusting method has the following problems. The visually unpleasant block distortion that stands out at the corners of the pixel block is
This is particularly noticeable when the pixel value of one of the corners of the four blocks at the intersection of block boundaries is larger than the pixel values of the other corners. However, with the conventional image quality adjustment method,
It is difficult to sufficiently improve the image quality because the processing is not performed with consideration for the corners of the pixel block. SUMMARY OF THE INVENTION The present invention provides an image quality adjusting method that solves the problem that the above-mentioned conventional technique has with respect to the point that the processing for the corners of the pixel block is not appropriate and that visually unpleasant distortion remains in the reproduced image. .

[0008]

In order to solve the above-mentioned problems, the first invention compresses an image signal from an original image using DCT conversion for each pixel block composed of N × M pixels. Then, for the image signal restored using the IDCT transform,
The following method is taken in the image quality adjusting method for removing the distortion caused by the compression and generating the image signal for reproduction. That is, the first invention is a pixel identification process for identifying that a pixel to be processed is a corner pixel in each block,
A detection process for detecting that the pixel at the corner is the pixel with the conspicuous distortion, and a filtering process for smoothing only the pixel value of the pixel with the conspicuous distortion to the pixel values in the vicinity of the pixel are sequentially performed. A second aspect of the present invention is the detection process according to the first aspect of the present invention, in which four corners are respectively in contact with a pixel block including the pixel to be processed and a portion where block boundaries of pixel blocks adjacent to the pixel block intersect vertically and horizontally. A two-pixel extraction process for extracting two pixels from the pixel, and a difference calculation process for calculating an absolute value of a difference between pixel values of the two pixels,
Repeating the two pixel extraction process and the difference calculation process,
Iterative processing for calculating the absolute value of the difference between pixel values for all combinations of the two pixels, comparison processing for comparing each result of the repeated processing with a set threshold value, and comparison processing for the comparison processing The determination process of determining whether or not the pixel to be processed is the pixel in which the distortion is conspicuous based on the result is performed.

[0009]

According to the first aspect of the invention, since the image quality adjusting method is configured as described above, it is identified by the pixel identification process that the pixel to be processed is a pixel at a corner in each block, and the detection process It is detected that the pixel at the corner is a pixel in which distortion is noticeable. Based on the detection result, when the pixel to be processed is a pixel in which distortion is conspicuous, the pixel value is smoothed with respect to the pixel values of neighboring pixels by the filtering process. Second
According to the invention, in the detection process of the first invention, the 2-pixel extraction process, the difference calculation process, and the iterative process are performed, and the block boundaries of the pixel block including the pixel to be processed and the pixel blocks adjacent to the pixel block are aligned vertically and horizontally. From the values of the four corner pixels that are respectively in contact with the intersecting portion, the absolute values of the pixel value differences for all combinations of two of the four pixels are calculated. The results and the set threshold value are compared in the comparison process.
Based on the result of the comparison processing, it is determined by the determination processing whether or not the processing target pixel is a pixel in which distortion is conspicuous. Therefore, the above problem can be solved.

[0010]

FIG. 1 is a flow chart showing an image quality adjusting method according to an embodiment of the present invention. In this image quality adjustment method, a digitized original image signal is compressed for each pixel block composed of N × M pixels by using DCT conversion, and is restored by using IDCT conversion. This is an image quality adjustment method for removing a distortion caused by compression to generate a reproduction image signal. The flowchart of FIG. 1 is for performing image quality improvement using, for example, a CPU (central processing unit) having a calculation unit, a control unit, and a register unit, and a storage device, and the following steps S210 to 270 are performed. It In the image quality adjustment method of FIG. 1, step S210 of storing the input image signal for each pixel in a storage device, and step S220 of identifying whether the input image signal is an image signal of a pixel at a corner of a pixel block. And step S230 of detecting from the image signal of the corner pixel of the pixel block whether or not the pixel is a corner pixel in which distortion is conspicuous. In the image quality adjustment method of FIG. 1, step S2 is further performed.
In step 40, in response to the result of step S230, filtering is performed on the image signal of the pixel of the corner where the distortion is noticeable. Also,
In step S250, the series of steps S210 to S210 described above.
The pixel to be processed in S240 is changed to the next pixel. A step S260 decides whether or not the image signal is finished, and a series of operations is finished in a step S270.

Next, the general flow of the image quality adjusting method of FIG. 1 will be described. In the image quality adjustment method of FIG. 1, the distortion of the reproduced image is removed and the image quality is improved by the following flow from (7) to (12). (7) Image signal storage process In step S210, the image signal for each pixel generated through the DCT conversion and the IDCT conversion is sequentially stored in, for example, a storage device. (8) Pixel Identification Processing In step S220, it is identified whether the image signal of the processing target pixel read from the storage device is the image signal of the pixel at the corner of the pixel block. This identification is, for example,
It is determined based on the storage address of the storage device. As a result, when the pixel to be processed is not the pixel at the corner of the pixel block, the process proceeds to step S250, and the series of processes moves to the next pixel. If the pixel to be processed is a pixel at the corner of the pixel block, the process proceeds to step S230. (9) Detection Processing When the processing target pixel is identified as a pixel at the corner of the pixel block, it is detected in step S230 whether or not the pixel is a pixel in which distortion caused by compression is conspicuous. When it is determined as a result of the detection that the pixel is not a pixel in which distortion is conspicuous, the process returns to step S220 of the pixel identification process again via step S250, and the process proceeds to the next pixel. On the other hand, when it is determined that the pixel to be processed is a pixel in which distortion is conspicuous, the process proceeds to step S240. (10) Filtering process In step S240, only the pixel values of the pixels in which the distortion is conspicuous among the corner pixels are filtered, and the pixel values in the vicinity of the pixels are smoothed. (11) Pixel change processing After the filter processing of step S240 is completed, step S240
250 is performed, the process returns to step S220 of the pixel identification process again, and the process proceeds to the next pixel. (12) Image signal end determination processing In step S260, it is determined whether or not the input of the image signal is completed. If it is determined that the processing has not ended, the process returns to step S220 of the pixel identification processing, and if it is determined that the processing has ended, a series of processing ends in step S270.

FIG. 6 is a diagram showing pixels used in the detection processing of this embodiment, and FIG. 7 is a flow chart showing the detection processing of this embodiment. Here, in FIG. 6, a part of the image to be processed is simplified and enlarged. In FIG. 6, circles drawn by solid lines or broken lines represent each pixel of the image to be processed, and solid lines drawn vertically and horizontally indicate boundaries of pixel blocks. The solid circles represent the pixels at the block boundary, and the dashed circles represent the other pixels. Next, the details of the above-described detection processing (9) will be described with reference to FIGS. 6 and 7. (9-1) Two-Pixel Extraction Process, Difference Calculation Process, and Iterative Process When the pixel A is identified as the pixel at the corner of the block as a result of step S220 of the pixel identification process, the step S231 in FIG. The three pixels B, C, and D closest to the point where the solid lines drawn vertically and horizontally like 6 intersect are obtained. Next, step S232 is implemented. Step S
232 sequentially extracts the pixel values of two pixels from the four pixels A, B, C, and D, and the calculation unit repeatedly performs the difference calculation of the formula for the extracted pixel values. α1 = | a−b | / MAX α2 = | a−c | / MAX α3 = | a−d | / MAX α4 = | b−c | / MAX ... α5 = | b−d | / MAX α6 = | c -D | / MAX where a, b, c, d; pixel value of each pixel A, B, C, D MAX: maximum value of pixel values a, b, c, d Here, for example, | a−b | represents the absolute value of the difference between the pixel value a and the pixel value b. (9-2) Comparison processing and determination processing Each difference absolute value α1 to α6 of the result of the formula obtained in step S232 is compared with the set threshold value TH by the CPU in steps S233 to 237, respectively. As a result of the comparison, it is determined whether or not the pixel to be processed is a pixel having a corner where distortion is conspicuous.

That is, in step S233, the absolute difference value α1 is compared with the set threshold value TH, and α1 ≧ TH.
In case of, it progresses to step S234. Step S23
In step 4, the absolute difference value α2 is compared with the set threshold value TH,
When α2 ≧ TH, the process proceeds to step S235. In step S235, the difference absolute value α3 is compared with the set threshold value TH, and when α3 ≧ TH, the process proceeds to step S236. In step S236, the absolute difference value α4 and the set threshold value TH are set.
Is compared, and when α4 <TH, step S2
37, the absolute difference value α5 is compared with the set threshold value TH in step S237, and when α5 <TH, the process proceeds to step S238. Finally, in step S238, the absolute difference value α6 is compared with the set threshold value TH, and α6
When <TH, it is determined that the pixel A has an angle where distortion is more noticeable than other pixels, and the process proceeds to step S240, and the subsequent processes are performed. Steps S233-S2
If even one of 38 does not satisfy the condition, the process proceeds to step S250, and the next pixel becomes the target pixel of the series of processes. Here, the process in step S230 will be described using specific numerical values. For example, each pixel A,
The pixels a, b, c and d of B, C and D are respectively a = 10.
0, b = 30, c = 25, d = 20, and the threshold TH is T
When H = 0.3, α1 = | 100-30 | /100=0.7 ≧ 0.3 α2 = | 100-25 | /100=0.75≧0.3 α3 = | 100-20 | / 100 = 0.8 ≧ 0.3 α4 = | 30-25 | /100=0.05 <0.3 α5 = | 30-20 | /100=0.1 <0.3 α6 = | 25-20 | /100=0.05<0.3, and the pixel value (100) of the pixel A is equal to another pixel value (3
0, 25, 20).

Next, details of the processing in step S240 will be described with reference to the drawings. FIG. 8 is a diagram showing the filter processing position of this embodiment, which is a simplified and enlarged part of the image. In FIG. 8, circles drawn by solid lines or broken lines represent pixels of the image to be processed, and solid lines drawn vertically and horizontally represent boundaries of pixel blocks. Solid line circles represent pixels at the block boundary, and dashed circles represent other pixels. FIG. 9 is a flowchart showing the filter processing of this embodiment. If it is determined in step S230 of the detection process that the corners are conspicuous, in step S241 in FIG. 9, the four pixels closest to the intersection of the solid lines drawn vertically and horizontally in FIG. 8 are A (pixel value a) and B.
(Pixel value b), C (pixel value c), and D (pixel value d). In FIG. 8, if the pixel value a of the pixel A is significantly different from the pixel values b, c, d of the other pixels, step S23.
The case where the judgment is 0 is shown. In such a case, the pixel block including the pixel A has a corner represented by a thick solid line in FIG. Therefore, in step S242 in FIG. 9, with respect to the pixel value of the peripheral pixel of the pixel A (for example, the pixel indicated by diagonal lines in FIG. 8) in the pixel block including the pixel A and the pixel value a of the pixel A, Filter processing is performed. As a result, the corners of the thick solid line in FIG. 8 caused by the pixel value a being significantly different from the other pixel values b, c, d are rounded. The filtering process is
For example, the average value is obtained using the peripheral pixels of the pixel A (including outside the pixel block), and the average value is set as the pixel value of the pixel A.
As described above, in the present embodiment, the distortion caused by the DCT transformation that occurs in the boundary portion of the pixel block is determined from the pixel values of the pixels at the corners of the pixel block, and only that pixel is compared with the peripheral pixels based on the determination result. Has been filtered. Therefore, it is possible to improve the image quality by removing the distortion that was impossible with the conventional technology.

The present invention is not limited to the above embodiment, and various modifications can be made. The following are examples of such modifications. (i) In the filtering process, the average value of the pixel values of the peripheral pixels and the pixel value a is obtained, and the average value is substituted for the pixel value a. For example, each pixel value of the peripheral pixels is weighted, A method of obtaining the average value of the weighted pixel values and the pixel value a is also conceivable. (ii) The threshold TH in the detection process is not limited to 0.3 and can be changed according to the application. (iii) One pixel (for example, pixel A) is set in one corner of each pixel block, and filtering is performed by determining whether the corner is a prominent corner or an inconspicuous corner. On the other hand, a plurality of pixels may be set, and a filtering process may be performed by determining whether a corner is conspicuous or an inconspicuous corner. As a result, it is possible to further improve the image quality.

[0016]

As described above in detail, according to the first aspect of the present invention, the distortion caused by the DCT conversion generated at the boundary portion of the pixel block is determined whether the pixel at the corner of the pixel block is a conspicuous corner or not. Detection is performed by the detection processing, and based on the detection result, only the pixels at the corners where distortion is noticeable are filtered to correct the distortion. Therefore, it is possible to improve the image quality by removing the distortion that is generated at the corners of the pixel block, which is not possible with the conventional technology. According to the second aspect of the present invention, the detection process is performed by selecting two pixels from the four corner pixels that respectively contact the pixel block including the pixel to be processed and the pixel blocks adjacent to the pixel block where the block boundaries intersect in the vertical and horizontal directions. 2 pixel extraction processing for extracting the pixels and the difference calculation processing for calculating the absolute value of the difference between the pixel values of the 2 pixels are repeated, and the pixel is calculated for all combinations of the 2 pixels. The absolute value of the difference between the values is calculated, and each result is compared with the set threshold value. Therefore, it is possible to easily and surely detect whether or not the pixel to be processed is a pixel having a corner with conspicuous distortion, for example, from a combination of predetermined results.

[Brief description of drawings]

FIG. 1 is a flowchart showing an image quality adjusting method according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a conventional image quality adjustment method.

FIG. 3 is a diagram showing pixels used in conventional edge detection processing and filter processing positions.

FIG. 4 is a flowchart showing a conventional edge detection process.

FIG. 5 is a flowchart showing conventional filter processing.

FIG. 6 is a diagram showing pixels used in the detection processing of the present embodiment.

FIG. 7 is a flowchart showing a detection process of this embodiment.

FIG. 8 is a diagram showing filter processing positions of the present embodiment.

FIG. 9 is a flowchart showing a filter process of this embodiment.

[Explanation of symbols]

S220 Performing pixel identification processing S230 Performing detection processing S231 to S232 Performing two-pixel extraction processing, difference calculation processing and repetition processing S233 to S238 Performing comparison processing and determination processing S240 Performing filter processing

Claims (2)

[Claims] 1. An image signal from an original image is N × M (N, M
Is a positive integer) For reproduction, by removing the distortion caused by the compression from the image signal compressed using the discrete cosine transform for each pixel block and restored using the inverse discrete cosine transform In an image quality adjusting method for generating an image signal, a pixel identification process for identifying that a pixel to be processed is a corner pixel in each block, and a detection for detecting that the corner pixel is the distorted pixel An image quality adjusting method, which comprises sequentially performing a process and a filtering process of smoothing only pixel values of a pixel in which the distortion is noticeable with respect to pixel values in the vicinity of the pixel. 2. The detection processing comprises two pixels out of four corner pixels which respectively contact a portion where the block boundaries of the pixel block including the processing target pixel and a pixel block adjacent to the pixel block intersect in the vertical and horizontal directions. A two-pixel extraction process, a difference calculation process for calculating an absolute value of a difference between pixel values of the two pixels, and a two-pixel extraction process and a difference calculation process.
Iterative processing for calculating the absolute value of the difference between pixel values for all combinations of the two pixels, comparison processing for comparing each result of the repeated processing with a set threshold value, and The image quality adjustment method according to claim 1, further comprising: a determination process of determining whether or not the processing target pixel is a pixel in which the distortion is conspicuous based on a result.