TW404136B - Process method preceding the image compression and the device itself - Google Patents

Abstract

The invention is a process method preceding the image compression and the device itself. It uses Just-Noticeable-Distortion (JND) technique to take account of the most notable between Contrast masking and Spatial masking effects to acquire a JND function value. It adjusts the brightness value of each pixel in the spatial domain towards the averages of the image block brightness, and the adjustment is under the JND function value of the pixel. The bit stream length is effectively shortened after the image compression by using the masking effects of human visual system. It reduces the perception redundancy related to the contrast masking effect without affecting the image quality to make the pixel brightness after the adjustment concentrated to the brightness average, so that the high frequency part of the image is deleted and the bit number required for image compression/decoding is reduced.

Description

File: TW0028D3 revl

File: TW0028D3 revl Printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs _B7__15. Invention Description M) 404136 The present invention relates to a method and device for image compression preprocessing, and in particular, to a method for processing discrete images Image compression preprocessing method and device designed by image cosine transform (DCT) -based image encoders (such as JPEG, MPEG, and H.263). According to the research of visual physiologists, there are two main Masking effects in the Human Visual System, which are briefly described as follows: (1) Contrast Masking The human visual system has different background sensitivity (Contrast Sensibility) to different background luminance, which means that the same image appears in different backgrounds and can be perceived by the human visual system to different degrees For example, an image in a black background will be harder to perceive than the same image in a white background, and an image in a white background will be less perceptible than the same image in a gray background. (2) Spatial Masking or Texture Masking 'This kind of masking effect mainly comes from the unevenness of the background brightness and spatial distribution. In fact, if the image changes occur at a high spatial frequency (Spatial Frequency) ) Background is less likely to be perceived by the human visual system than at low spatial frequency backgrounds. Therefore, when a background brightness has a considerable degree of spatial distribution frequency, an image can be placed in or taken out of this background, which cannot be detected at all. Perceived by the human visual system. The basic principles of this image compression (Image Compression) pre-processing method (please read the precautions on the back before filling this page). _ Order) A4 size (210X297 mm) Printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs

File: TW0028D3_revl A / B7 ------, 40ΓίΤ3ΐ --- V. Description of the invention (2) Based on the above-mentioned masking effect, the system is used to 'remove parts that cannot be distinguished by the human visual system during image compression and encoding. 'For example: an image with a high spatial frequency background or an image with low brightness contrast sensitivity' can reduce the memory space occupied by image compression without affecting the image quality. Before image compression, the 'excessive use of memory space caused by stored images' is called Perceptual Redundancy. 0 The traditional method for quantitatively estimating the above-mentioned perceptual redundancy is based on the use of the correct method proposed by Jayant, Johnston, and Safranek. Just-Noticeable-Distortion (JND) function 値, this JND function 値 is a function defined on the spatial domain (Spatial Domain), it represents a threshold 値 (Threshold) of the brightness change of each spatial location, such as Formula (1) shows: JND (x, y) = max {/ 1 (6 (^, y), m (x, y) \ f2 (b (x, jf))} (1) where b (x , Y) represents the background brightness near Pixel (x, y) Mean 値, m (x, y) represents the average brightness change 附近 near the pixel (x, y), while the function f \ (x, y) simulates the above-mentioned spatial shadowing effect, and the function f2 (X) simulates the above-mentioned contrast Masking effect. In addition, the meaning of taking maximum 値 (max) in formula (1) is to take out one of the above two masking effects in the same pixel (X, y). When the image is compressed and encoded, if each The brightness change of each pixel is lower than the JND threshold. The reduction in image quality during the compression process cannot be detected by the human visual system. Please refer to Figure 1, which shows the traditional DCT-based removal of spatial occlusion— = W-^ — 装-(Please read the precautions on the back before filling this page)

.1T

T .Λ This paper size applies to China National Standards (CNS) A4 (210 X 297 metric) printed by the Central Government Standards Bureau of the Ministry of Economic Affairs for consumer cooperation

File: TW002 8D3 revl " B7 V. Description of the invention 0) 404136 Block diagram of perceptually redundant image coder (Image Coder) caused by shadowing effect. The input image is first divided into 8X8 image blocks (Image

Block) IB, and then DCT-converts the image block IB using the DCT processing unit 11 and outputs 64 DCT conversion coefficients TC to the Quantization Unit 13 and the Quantization Matrix Unit.

Matrix Unit) 15, the quantization unit 13 then uses a quantization matrix provided by the quantization matrix unit 15 to quantize the DCT conversion coefficients TC and output a group of quantization coefficients QC to a variable-length encoding unit (Variable-Length

Coding Unit) 17 'is used to encode the quantization coefficient QC. Finally, a bitstream BS is output.

The quantization matrix unit 15 generates a group of parameters based on the 64 DCT conversion coefficients TC. For different image encoders, the number and size of the group parameters have different upper and lower limits, and different neighboring parameters. Gap 値 For example, for JPEG and MPEG, this group of parameters has a total of 64 ', which constitutes an 8X8 quantization matrix, while H.263 uses only one parameter. The above-mentioned image encoder is a perceptual measurement for the above-mentioned DCT conversion coefficient TC, that is, setting parameters in the quantization matrix, so that the quantization coefficient QC generated by the quantization unit 13 belongs to the high-frequency part. The coefficient of is set to 0, thereby reducing the number of bits of the bit string BS (in general, the more coefficients in the quantization coefficient qC are 0, the shorter the length of the bit string BS) to improve the compression effect. Although the above method has the function of bit reduction (Bit Reducing) & degree is applicable to the Chinese Grandfather-(Please read the precautions on the back before filling this page). Install. • 'Order ---- J-- "trip

File: TW〇〇28D3_ A7 B7 Printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs 5. Description of the invention (called 404136), but still has the following disadvantages: 1. Because each image encoder has a specific quantization matrix arrangement method (Coding Syntax), such as the above-mentioned upper and lower limits, and the gap between adjacent parameters. Therefore, if the method of setting a quantization matrix is used to achieve the purpose of eliminating perceptual redundancy, the parameters of the quantization matrix must be different with different encoders. Make changes to it, which increases the complexity of the system. 2. Because the DCT conversion coefficient is measured by the above perception, the spatial frequency characteristics of the input image are processed (excluding the high frequency part), and the brightness of each pixel in the image is not considered. Variation characteristics, such as the difference between different parameters in the quantization matrix 因此, therefore, the perceptual redundancy caused by the above-mentioned contrast masking effect cannot be eliminated by this method. In view of this, the main purpose of the present invention is to provide an image compression Pre-processing device, which is located in front of the DCT-based image encoder, and is used to pre-process the image. The adjustment of pixel brightness and image compression and coding can not only eliminate the perceptual redundancy of the image, but also do not need to adjust different quantization matrix parameter setting methods with different image encoders to simplify the operation of the system. Another object of the invention is to provide an image compression pre-processing method, which uses the above-mentioned JND function 値 as a basis for adjusting pixel brightness ， and adjusts the brightness 每个 of each pixel in the image to reduce the perception caused by space and contrast occlusion effects. Redundancy, without affecting the quality of the image, can further reduce the number of bits required for image compression. According to the main purpose of the present invention, an image compression pre-processing device is provided. The device includes: This paper scale is applicable to the country of China Standard (CNS) A4 specification (210X297 mm) (Please read the notes of C & fill in this booklet. Ir, spring A7 B7

File: TW〇〇28D3 re, V. Description of the invention (Ai) 404136 A block division unit is used to divide the input image into 8X8 image blocks; a JND calculation unit is used to calculate each image based on the above images The prime JND function 値 forms a JND chart (Profile); the distortion masking unit (Distortion Masking Unit) receives the above-mentioned image blocks and JND charts for the brightness of each pixel in the image block 値, Make adjustments to the brightness average of this image block; and a block synthesis unit that receives the output of the above-mentioned distortion masking unit, combines the image blocks into a new image, and sends it to a DCT-based image encoder for compression and Encoding (eg: iPEG 'MPEG, H.263) to form a bit string. The distortion masking unit uses this JKD chart to adjust the brightness of each pixel in the image block, and the average brightness of 1¾¾, which can eliminate the verbosity of the human visual system and the device is set in front of the image encoder, so there is no need to change it. The image encoder's internal settings eliminate perceptual redundancy. According to another object of the present invention, a method for pre-compressing an image is proposed, the steps of which are briefly described as follows: • First 'divide the image into 8 × 8 image blocks; then calculate the jND function for each pixel in the image block. To form a JND chart; according to this JND chart, obtain weighted (Weighted) numbers for the JND functions; obtain the average brightness of each image block; and according to these brightness averages and weighting functions, , Adjust the brightness of all pixels in each image block 値 'toward the average brightness of the image block, and form a new image block; finally, combine the new image blocks into a new image for compression And encoding. This method reduces the contrast of the brightness of adjacent pixels (that is, the difference in brightness 缩小 is reduced).

CNS) Ege (2i〇x ^ ¥ T (Please read the precautions on the back before filling this page)

, 1T Printed by the Consumer Cooperatives of the Central Bureau of Standards of the Ministry of Economic Affairs

File: TW0 0 2 8 D3 re, A7 B7 Printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs 5. Description of the invention (0 reduces the above two types of perceived redundancy, reduces the number of bits required for compression coding, and is adjusting When the pixel brightness of different image blocks is 値, a weighting function 求 is obtained to maintain the quality of the combined image blocks. In order to make the above-mentioned objects, features, and advantages of the present invention more obvious and easier to understand, one of the following is specifically enumerated. The preferred embodiment and the accompanying drawings are described in detail as follows: Brief description of the drawings: Figure 1 shows a block diagram of a traditional image encoder based on the DCT to remove the perceptual redundancy caused by the spatial shadowing effect. Figure 2 is a structural diagram of an image processing device according to a preferred embodiment of the present invention; Figure 3 is a structural diagram of a distortion masking unit in Figure 2; Figure 4 is a diagram of Figure 3 Structure diagram of the middle adjustment unit; Figure 5 shows the schematic diagram of adjusting the brightness of pixels with the structure of Figure 4; Figures 6A ~ 11A show the traditional JPEG and H.263 image maps in sequence; and 6B ~ 11B are not drawn in sequence relative to 6A ~ 11A The JNDM image of the Chinese image processed by the present invention. Symbol description = 11: DCT processing unit 15: quantization matrix unit 21: block segmentation unit 25: distortion masking unit 13: quantization unit Π: variable length coding unit 23: JND calculation Unit 27: Block Synthesis Unit (please read the precautions on the back before filling this page)

、 1T Λ The size of this paper is applicable to the Chinese National Standard (CNS) A4 specification (210 × 297 mm) le: TW0028D3 revl A7 B7 V. Description of the invention (7) 404136 29: DCT image encoder 31 33: Weighting matrix unit 35 41: Comparison unit 43 45: Subtraction unit 47 49: Extremely small unit Average 値 Calculation unit Adjustment unit Addition unit Extremely large unit Printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economics A preferred embodiment Please refer to FIG. 2, which shows a comparison according to the invention Structure of an image compression processing device of the preferred embodiment. This image compression processing device includes: a block division unit 21 for dividing the input image frame (Frame) F into a plurality of non-overlapping (Ncmoverlapping) 8X8 image blocks IB; a JND calculation unit 23 for performing For the image frame F, calculate the JND function for each pixel. Here, JNDj⑴ is used to represent the JND function i of the i-th (i = l ~ 64) pixels in the j-th image block IB to form a JND chart J The distortion masking unit 25 receives the above-mentioned image block IB and JND chart J, and is used to adjust the brightness of each pixel in the image, and outputs an image block and a block synthesis unit 27 to receive all the above-mentioned images The blocks are combined into a new image frame Fjndm, which is then transmitted to the DCT image encoder 29 for compression and encoding to generate an image bit string BS. Please refer to FIG. 3, which shows the structure of the distortion masking unit 25 in FIG. Its structure includes: an average 値 calculation unit 31 for obtaining a brightness average 値 for each image block, where the brightness average 値 of the first image block IB is represented here; a weighted matrix unit (please refer to Please read the notes on the back before filling this page.) Binding. Order '.4. This paper size is applicable to China National Standard (CNS) A4 (210X 297 mm)

File: TW0 02 8 D3 revl A / B7 5. Invention Description (s) 33. According to the JND chart J above, it is used to weight the JND function 每个 for each pixel. Here, JND7 (0 represents the jth A weighted JND function i of the i-th pixel in the image block IB; and an adjusting unit 35 for adjusting the brightness of each pixel in the image according to the brightness average decay and the weighting function 値 "(〇, (0, the adjusted pixel brightness 値 is represented by / fDM (0, as shown in formulas (2), (3), and (4). \ Msx {A, Ρ} \ if Pj (J) > Pj , [min {^ (, Pj \ otherwise, (2)

A

Pj (i) + B (3) ---------- ^ 装 -I (Please read the precautions on the back before filling this page)

B |-1, iiPj (i) > Pj, [+ 1, otherwise, (4) where JNDf.JNDjQxWMQ, WM (i) (i = l ~ 64) represents the weighted 値 of the i-th JND function 値, where The 64 weighted 値 can form a matrix form, called a weighting matrix, as shown in Table 1. Table 1 0.4 0.8 0.4 0.8 0.4 0.8 0.4 0.8 0.8 1 1 1 1 1 1 1 0.4 0.4 1 1 1 1 1 1 0.8 0.8 1 1 1 1 1 1 0.4 0.4 1 1 1 1 1 1 0.8 0.8 1 1 1 1 1 1 0.4 0.4 1 1 1 1 1 0.8 0.8 0.4 0.8 0.8 0.4 0.8 0.4 0.8 0.4 Among them, the pixels located on the four boundaries of each image block IB after F segmentation of the daytime image are weighted by 値 (both 0.4 and 0.8) Cross-distribution) This paper size applies to China National Standard (CNS) Α4 specification (210X297 mm)

, 1T Printed by the Consumer Cooperatives of the Central Bureau of Standards of the Ministry of Economic Affairs

File: TW0028D3 revl A7 B7 Printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs 5. Description of the invention (1 ^ ~~ 4ϋ4ΐ3ί, ..., the gain of information is lower than the weighted 値 (1.0) of other parts, so the boundary map can be reduced The threshold of the prime JND is 値, and the brightness of the pixel will be adjusted. When the 8 X8 image block IB is combined into a new image frame F, the artificial blocking effect caused by the recombined image can be reduced. Please refer to Fig. 4, which shows the structural diagram of the adjustment unit in Fig. 3. Its structure includes: a comparison unit 41 for comparing the brightness of each pixel 値 値 and the average brightness 値 Λ: addition unit 43 Is used to compare the pixel brightness smaller than the mean value after comparison by the comparison unit 41 (0 ′ plus a weighting function 値 JNE> 7 (〇; subtraction unit 45), which is used to compare The pixel brightness 値 of the average 値 Λ is (0 'minus a weighting function 値 JND; T (0; a maximum unit 47 for receiving the output 値 of the subtraction unit 45, which is compared with the brightness average 値, and which is Larger is the new pixel brightness, ρΓΜ (0; and A small unit 49 is used to receive the output 加 of the adding unit 43 and compare it with the brightness average 値, and take the smaller 値 as the new pixel brightness 値 Please refer to FIG. 5, which shows that it is better according to the present invention. Schematic diagram for adjusting the pixel brightness 値 by the adjustment unit in the embodiment. The white circle 圏 represents the unadjusted pixel brightness 値 is (0, the black circle represents the adjusted pixel brightness 値 PfDM (0, and the double arrow length represents the pixel brightness値 is the size that can be adjusted ', which is the weighting function of the pixel (IND%). From Figure 5, it can be clearly seen that the adjusted pixel brightness (shown by the black circle) is less than the unadjusted pixel. The brightness 値 ρχ〇 (shown by the white circle) is more concentrated on the average brightness 値 Λ (shown by the thick black line in the middle) of the image block IB, and each adjusted pixel brightness 値 if'DM (0 and the original The pixel brightness is equal to (0), the absolute paper size is applicable to the Chinese National Standard (CNS) M specifications (2 丨 0 > < 297 mm I -----,-installation ------ ^ order ί ----- ^ Caution (Please read the notes on the back before filling in this page)

File: TW0 0 2 8 D3 re, A7 B7 V. Description of the invention (I.) 404136 The confrontation must not be greater than the clock, as shown in formula ⑺. Because shouting 0 値 must not be larger than the foot 根据, and according to the definition of _ _⑺, the formula ⑺ indicates that the change in image quality after the present __brightness_ is invisible to the human visual system. | ^ DM (〇- ΡΛή \ (0 · (5) Printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs, please refer to Figures 6A ~ 11A and Figures 6b ~ Ub. Among them, Figures 6A ~ 8A represent aircraft and baboons, respectively , JPEG image maps of family portraits; Figures 9A-11A represent the H 263 image maps of virtual animals, mother and daughter, and broadcasters; Figures 6B-11B represent the aforementioned jpEG images and h.263 images processed by the present invention. Please refer to Table 2 and Table 3 at the same time, and compare the list of bits required in the image using the present invention with the traditional JPEG image and H.263 image, respectively. From Table 2 and Table 3, we can know that for different JPEg The image and H 263 image are compressed / encoded. Compared with the original image, the obtained JNDM image is 'up to 9.27% ~ 14.49% Bit Reducing Ratio' without affecting the quality of the image. From the image diagrams shown in ~ 丨 1A and 6B ~ 1IB, it can be found that the processed JNDM image, the original JPEG image, and the H.263 image are almost different in human visual effects. Out, meaning the effect is almost the same, but Save a lot of storage bit rate. This paper scales applicable Chinese National Standard (CNS) A4 size (210X297 public Chu) (Please read the notes and then fill in the back of this page). Xu · -V. '.4

File: TW0028D3_revl B7 V. Explanation of the invention (") 404136 Table 2 JPEG image original image JNDM image Bit reduction rate (%) Bit number Bit number Plane 6161 5460 11.38 Baboon 9457 8572 9.36 Family portrait 6863 6158 10.27 J 1 — .. Table 3 -------- H.263 original image JNDM image bit number subtraction bit number bit rate (%) ... virtual animal 6244 5339 14.49 mother and daughter 15766 13913 11.75 broadcaster 8680 7875 9.27 Therefore, the present invention has the following advantages: Printed by the Consumer Cooperative of the Central Bureau of Standards of the Ministry of Economic Affairs (1) Because the image compression pre-processing device of the present invention is installed before the image encoder, the pixel brightness of the image is first After the adjustment, the result is passed to the image encoder for compression and encoding. Therefore, it is not necessary to adjust the parameter setting method of the image encoder with different image encoders, which can simplify the operation process of the system. (2) Use the JND function 値 to adjust the brightness 每 of each pixel toward the brightness average 影像 of the image block, so that the adjusted pixel brightness 値 is more concentrated on the brightness average 値 (usually the low frequency part), which can be removed The high-frequency signal part of the image reduces the number of bits after image compression / encoding. Specifications of this paper Gu t @ S 家 张 准 （CNS） A4 Specification （

File: TW0028D3 rev1 A / ___ ~ ___ B7___ V. Description of the invention (π) 404136 (3) Use the JND technology to adjust the brightness of each pixel in the spatial area, and adjust it toward the average brightness of the image block, and adjust the The JND function whose size does not exceed the pixel (as shown in the formula), the jND function does not consider those with significant contrast and spatial shadow effects. Therefore, it can reduce the perceptual redundancy related to the contrast shadow effect at the same time, without affecting the image. Under the premise of quality, the number of bits required for image compression / encoding is further reduced than the conventional technology. In summary, although the present invention has been disclosed as above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Retouching, so the scope of protection of the present invention shall be determined by the scope of the attached patent application. (Please read the notes on the back before filling out this page) Binding Printed by the Staff Consumer Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs

Claims (2)

Printed by A8 File: TW0 0 2 8 D3_revl 〇8 ~ D8 of the Central Standards Bureau of the Ministry of Economic Affairs 6. Application for patent scope 1. An image compression pre-processing method for pre-compressing a first image. The first image includes a plurality of pixels, and the steps include: dividing the first image into a plurality of first image blocks; and based on the first image, calculating the pixels @ plural just-perceivable distortion (Just -Noticeable-Distortion; JND) function 値; according to the JND functions 値, adjust the complex numbers of the pixels in each of the first image blocks based on a brightness average 値 of each of the image blocks And a plurality of second image blocks; and combining the second image blocks into a second image. 2. The image compression pre-processing method described in item 1 of the scope of the patent application is an image compression pre-processing method that uses discrete cosine transform (DCT) for image coding. 3. The image compression pre-processing method as described in item 1 of the scope of the patent application, wherein the first and second image blocks are non-overlapping image blocks. — — — 4. The image pre-compression processing method described in item 1 of the scope of the patent application, wherein the JND function 値 is a function 定义 defined on a spatial region, and one of the pixels corresponding to the JND functions 取 is taken. The maximum number of functions ® and i i i i i i i i i i i i i i i i i i 拟 i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i had had had since since since since when if if if i if if if if if since since since been the case since then since when since then since when since then if any-since the number i and i have been the same 5. For the image compression pre-processing method described in item 4 of the patent application park, (please read the precautions on the back before filling this page) This paper size applies to Chinese National Standard (CNS) A4 specification U10X297 mm. Printed by A8 B8 File: TW0 0 2 8 D3 revl C8 " " D8 of the Central Standards Bureau of the Ministry of Economic Affairs. The function of the simulated spatial shadowing effect is obtained from the average brightness of the scene near each of these pixels, and an average of the brightness change, and / or 3 The function of the simulated contrast shadowing effect is obtained from the vicinity of each of these pixels. The background brightness averages 値 _. 6. The image compression pre-processing method as described in item 4 of the scope of patent application, wherein the JND function 値 represents a threshold 値 for each of the first brightness changes. 7. The image compression pre-processing method as described in item 4 of the patent application, wherein the steps of adjusting the first brightness according to the JND function include: obtaining the brightness for each of the first image blocks. Average 値; obtain corresponding weighting functions 对该 for the JND functions 値; and adjust the first brightness 中 in each of the first images according to the brightness averages 値 and the weighting functions 値 to obtain Corresponding second brightness 値, and each of the second image blocks is formed. 8. The image compression pre-processing method as described in item 7 of the scope of patent application, wherein each of the weighting functions 値 is generated by multiplying each of the JND functions 上 by a weighting 値. 9. The image compression pre-processing method as described in item 8 of the scope of the patent application, wherein the weighting 値 is a fixed 该 for these weeks in the same first image block. 10. The image compression pre-processing method as described in item 8 of the scope of the patent application, wherein the weighting 値 is different from the pixels on the four boundaries of the first image block, which are smaller than 1.分布 Cross-distributed, and here (please read the notes on the back before filling out this page). Install-• 1C 1 · ί I (nm ^ ii Hal) This paper applies the Chinese National Standard (CNS) A4. 210X297 mm) File: TW0 0 2 8 D3 revl A8 Βδ C8 D8 The weight of the pixels printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs, District I, Area 1.0 is taken as 1.0. 11. If item 10 of the scope of patent application The image compression pre-processing method, wherein the weighting 値 relative to the pixels on the four boundaries of the first image block is taken as 0.4 and 0.8 値 cross-distribution, and other pictures in the second The weighting factor of the prime is taken as 1.0. 12. The image pre-compression processing method described in item 7 of the scope of the patent application, wherein the first brightness values are adjusted according to the brightness average values and the weighting functions. The steps include: ^ for each of the first image blocks The first brightness 値 is compared with each of the brightness averages 根据; according to the comparison result, the weighted functions 加上 are added to the first brightness 小于 smaller than the brightness averages 以 获得 to obtain a plurality of third brightness Yu Lou ^ and for the first brightness 大于 greater than the average brightness 各 of each te, subtract the weighting functions to obtain a plurality of fourth brightness 値;., .1, and the third brightness 値, Compare with each of the brightness averages ，, and take the smaller 値 as the second brightness 値, and compare the fourth brightness 値 with each of the brightness averages 取, and take the larger 値 as the The second brightness 値. 13. As described in the image compression pre-processing method described in item 7 of the scope of patent application, wherein the difference between the second brightness 値 of the pixels and the first brightness 値 is definitely not greater than the Some JND functions 値. 14_ The image pre-compression processing method described in the seventh fill of the patent application scope, wherein the adjusted second brightness in the first image block 値 (Please read the precautions on the back first (Fill in this page again) Binding J Paper Size Printed in Chinese National Standard (CNS) A4 specification (210X297 mm) ~ ----- .le: TW0028D3 rev: A8 B8 C8 D8 Printed by the Consumer Cooperatives of the Central Bureau of Standards of the Ministry of Economic Affairs 6. The scope of patent application is 404136. 19. Such as The image compression method described in the scope of the patent application No. π, wherein the weighting 値 relative to the pixels on the four boundaries of the first image block is taken as 0.4 W-0.8 two kinds of cross-distribution, and The weights of the other pixels of the first image block are taken as 1.0, and are used to adjust the block effects of the first images and the second image of the edge of the table image block. 20. The image compression method according to item 15 of the scope of patent application, wherein the step of adjusting the first brightness 値 includes: the first brightness 値 and the brightness in each of the first image blocks. The average 値% is relatively good. According to the comparison result, the second brightnesses that are less than each of the brightness averages are added to the weighting function 値 2 to obtain a plurality of third, brightness 値, and The second brightness 値 of the brightness average 値, subtracting the weighting functions 値 to obtain a plurality of fourth brightness 値; 苡 and comparing the third brightness 値 with each of the brightness distortions 値, Take the smaller 値 as the second brightness 値, and compare the fourth brightness 与 with the average brightness 値, and take the larger one as the second brightness 値 ^ 21. Application scope The image compression method according to item 15, wherein the differences between the second luminances 値 of the pixels and the first luminances 値 are absolutely not greater than the JND functions 値. 22. The image compression method described in item 15 of the scope of patent application, wherein each mu > B- —-— Bn— In i nn mV ·-II (Please read the notes on the back before filling in this page j- ------ This paper size is applicable to Chinese National Standard (CNS) A4 (210 X 297 mm) er ABCD Printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs 6. The scope of patent application for these first image blocks After the adjustment, the difference between each of the second brightnesses 値 and the brightness average 値 is definitely smaller than the difference between each of the nested brightness 値 and the ~ _ _ brightness average 値. Absolutely 値. 23 · —an image compression device, It is used to process a first image 'and compress it into a first bit string. The first image includes a plurality of pixels. The image compression device includes: ~ a segmentation unit for segmenting the first image Into a plurality of first image blocks; and a calculation unit for calculating a plurality of precisely detectable distortion (JND) functions 该 of the pixels based on the first image, and forming a graph for output; a mask A unit for receiving the first image blocks and the J The ND circle table uses a plurality of first brightness values 对该 of the first image blocks f to adjust an average brightness value of each of the first image to image blocks, and outputs a plurality of second image blocks. ; And a synthesis unit that receives the second images 1 ^ yi_ 丄 and combines them into a second image, which is then compressed and encoded to an image encoder based on discrete cosine transform to generate the ft — Bit string \ v _ 24 · The image compression as described in item 23 of the patent application range, wherein the plurality of first and second image blocks are non-overlapping image blocks. 25. The image compression device according to item 23, wherein the JND function is a function defined on a spatial region, and takes the first function that simulates the effect of spatial shadowing and a simulated contrast shadowing (please read the precautions on the back first) (Fill in this page again) • Packing --- Order J This paper size is applicable to Chinese National Standards (CNS) A4 specifications (210X297 mm) A8 B8 C8 D8 File: TW0 0 2 8 D3 re1-; ~~ 4041¾6 6. Apply for a patent The largest in a second function of the range 26. The image compression device as described in item 25 of the application scope, wherein the first function 値 is obtained from a background brightness average 1 and a brightness change average 附近 near each of the pixels. 27. 如 串The image compression device described in item 25 of the patent scope, wherein the second function 値 is obtained from the background brightness average 附近 near each of the pixels. 28. For example, the image compression device described in item 25 of the patent scope Where the JND function 値 represents a critical value of the first brightness changes. 29. The image compression device described in item 23 of the scope of patent application, wherein the masking unit includes: an average chirp calculation unit for obtaining a brightness average chirp for each of the first image blocks; a weighting matrix unit To obtain a plurality of weighting functions 根据 according to the JND function 値; and to integrate Qin 'yuan to receive each of the brightness averages 値 and the weighting functions Straight to adjust the The first brightness levels are obtained to obtain a plurality of second brightness levels, and the second image blocks are output. _ 30. The image compression device described in item 29 of the IcT patent application range, wherein the adjustment unit includes: = comparison_ unit, for comparing the brightness of the pixels in each of the first image blocks with each of the Brightness average 値 1 comparison; an addition unit for each of the first image blocks, after the ratio (please read the precautions on the back before filling this page)-installed to the central of the JND Ministry of Economic Affairs Printed by the Bureau of Standards Consumer Cooperatives The paper size applies to the Chinese National Standard (CNS) A4 specification (210X297 mm) Printed by the Consumers' Cooperative of Central Standards Bureau of the Ministry of Economic Affairs File: TW0028D3_revl °° _; _I 4Π4136 Shen_ VI. Patent application The difference between the ambient brightness and the average brightness 値 is definitely smaller than the difference between each of the first brightness 値 and the brightness average 値. 15. An image compression method for compressing a first image, the first image including a plurality of pixels, the steps include: _ cutting the first image ¥ into a plurality of first image blocks; according to the first An image, calculate a plurality of precisely detectable distortion (JND) functions 该 of the pixels to obtain a JND chart; according to the JND chart, for these: TND function 値 find a plurality of weighting functions 値 ·; Take a brightness average 値 of each of the first image blocks; *,-according to the brightness averages 値 and the weighting functions 値, for the plurality of first pixels of each of the first image blocks, A brightness level, adjusting towards a brightness average level of each of the first image blocks to obtain a plurality of second brightness levels and forming a plurality of first blocks; and combining the second image blocks into a first Two images, and then use the second image to perform compression and encoding. '16. The image compression method according to item 15 of the scope of patent application, wherein the plurality of first and second image blocks are rectangular image blocks which do not overlap. Π. The image compression method according to item 15 of the scope of patent application, wherein each of the weighting functions 値 is generated by multiplying each of the jND functions 上 by a weighting 値. 18_ The image compression method as described in item 17 of the scope of patent application, wherein the weighting is a fixed paper size for the pixels of the same first image block and applies the Chinese National Standard (CNS) A4 specification ( 2 丨 〇X 297 mm) (Please read the notes on the back before filling this page) 404136 .e: TW002 8D3 revl A8 B8 C8 D8 6. After comparing the patent application scope with the units, the first brightness 値 smaller than the respective brightness average 各 is added to the weighting functions 値; Method for comparing the first-brightness values greater than the average brightness value of each of the first image blocks after the comparison unit is subtracted from the weighted ancestors itJ_ 一 _ > A maximal unit that receives the output of the subtraction unit and compares it with each of the average brightness values' take the larger one as the second brightness value; ... " " '«~' and An extremely small unit receives the output of the addition unit for comparison with each of the brightness average 値, and takes the smaller 値 as the second brightness 値. 31. The image compression device described in item 23 of the scope of patent application, wherein the difference between each of the second brightness 値 of each pixel and each of the first brightness 1 is not greater than each of the JND functions. value. 32_ The image compression device according to item 23 of the scope of patent application, wherein the difference between the adjusted second luminances 値 and the average luminance 中 in each of the first image blocks is absolutely smaller than each of the first The difference between a brightness 値 and the average brightness 値 is absolutely 値. (Please read the precautions on the back before filling out this page. 1'Installation 丨 '1T isj-I— fm 4i Printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs »^ Jn- —HI— The paper scale is applicable to Chinese national standards ( CNS) A4 specification (210X297 mm) I ..... «ms tm« ^ i · —