Image processing system, image processing method, and non-transitory computer program product

Abstract

An image processing system, an image processing method, a non-transitory computer readable recording medium of a memory program, and a non-transitory computer program product are disclosed, wherein the image processing method comprises: receiving an input image; obtaining a front mask image of the input image and enlarging the front mask image to obtain a mask image, wherein a size of the mask image is the same as a size of the input image; performing an image quality function on the input image via an image quality module to obtain an image quality image; and performing a pixel weighted average procedure on the input image and the image quality image based on a value of each of a plurality of pixels of the mask image to obtain an output image.

Description

Technical Field

[0001] This invention belongs to the technical field of image processing, specifically techniques for local optimization of images. Background Technology

[0002] In multi-media chip architectures, a series of image quality (PQ) functions are used to improve image quality, such as noise reduction, sharpness adjustment, and decontour. With the advancement of display technology and the widespread adoption of 4K and 8K TVs, there is a technical need for more precise image quality functions, that is, the ability to perform appropriate processing on local areas of the image. However, traditional approaches typically perform general processing on the entire image, offering very limited local processing capabilities. To achieve specific range or local processing, in addition to adding conditional checks, these conditions usually require calculations on more pixels, as well as additional circuitry (such as line buffers) in the data path and control logic. Because each pixel needs to be processed individually in each clock cycle, the data path design becomes more complex, and the control logic becomes more cumbersome, leading to an increase in circuit area. Summary of the Invention

[0003] In view of the above, some embodiments of the present invention provide an image processing system, an image processing method, a non-transitory computer-readable recording medium for memory programs, and a non-transitory computer program product to improve the problems of the prior art.

[0004] Some embodiments of the present invention provide an image processing system, the image processing system including a processing module and an image quality module; the image quality module is configured to perform an image quality function; and the processing module is configured to receive an input image and perform the following steps: (a) obtaining a pre-masking image of the input image and magnifying the pre-masking image to obtain a masking image, wherein the size of the masking image is the same as the size of the input image; (b) performing an image quality function on the input image via the image quality module to obtain an image quality image; and (c) performing a pixel-weighted averaging procedure on the input image and the image quality image based on the value of each of a plurality of pixels of the masking image to obtain an output image.

[0005] Some embodiments of the present invention provide an image processing method executed by a processing module. The image processing method includes: (a) receiving an input image; obtaining a pre-masking image of the input image and enlarging the pre-masking image to obtain a masking image, wherein the size of the masking image is the same as the size of the input image; (b) performing an image quality function on the input image via an image quality module to obtain an image quality image; and (c) performing a pixel-weighted averaging procedure on the input image and the image quality image based on the value of each of a plurality of pixels in the masking image to obtain an output image.

[0006] Some embodiments of the present invention provide a computer-readable medium for a memory program and a non-transitory computer program product, which can perform the aforementioned image processing method after the processing unit loads and executes the program.

[0007] Based on the above, some embodiments of the present invention provide an image processing system, an image processing method, a non-transitory computer-readable recording medium for a memory program, and a non-transitory computer program product. By masking an image to obtain local information of an input image, and then using the local information of the input image to perform a pixel-weighted averaging process on the input image and the image quality image, the system can achieve the effect of performing specific image quality functions on a local area of ​​the input image without modifying the original architecture (e.g., circuit architecture) of the image quality module. When the image quality module is implemented in hardware, no additional circuit area is required without modifying the original architecture of the image quality module. Attached Figure Description

[0008] Figure 1 This is a block diagram of an image processing system according to an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating an image processing method according to some embodiments of the present invention. Figure 3 This is a block diagram of an image processing system according to an embodiment of the present invention. Figure 4A This is a schematic flowchart of a pixel-weighted averaging procedure drawn according to some embodiments of the present invention. Figure 4B This is a schematic diagram of a differential image drawn according to some embodiments of the present invention. Figure 5 This is a schematic diagram illustrating an image processing method according to some embodiments of the present invention. Figure 6 This is a block diagram of an electronic device system illustrated according to some embodiments of the present invention. Figure 7 This is a flowchart illustrating an image processing method according to some embodiments of the present invention. Figure 8This is a flowchart illustrating an image processing method according to some embodiments of the present invention. Figure 9 This is a flowchart illustrating the front shielding image setting process according to some embodiments of the present invention. Figure 10 A flowchart illustrating an image processing method according to some embodiments of the present invention. Figure 11 A flowchart illustrating an image processing method according to some embodiments of the present invention. Figure 12 A flowchart illustrating an image processing method according to some embodiments of the present invention. Detailed Implementation

[0009] The foregoing and other technical contents, features, and effects of this invention will be clearly presented in the following detailed description of embodiments with reference to the accompanying drawings. Any modifications and alterations that do not affect the effects and objectives achieved by this invention should still fall within the scope of the technical contents disclosed in this invention. In all the drawings, the same reference numerals will be used to denote the same or similar components.

[0010] Figure 1 This is a block diagram of an image processing system according to an embodiment of the present invention. Please refer to [link / reference]. Figure 1 The image processing system 100 includes a processing module 101 and an image quality module 102. The processing module 101 is configured to obtain an input image 103. The image quality module 102 is configured to perform an image quality function on the image input to the image quality module 102 based on an image algorithm to adjust the image input to the image quality module 102. The aforementioned image quality function includes, for example, noise reduction, decontour, contrast adjustment, color adjustment, and sharpness adjustment. The image quality module 102 can be implemented using circuitry or software.

[0011] The following describes in detail, with reference to the accompanying drawings, some embodiments of the image processing methods of the present invention and how the various modules of the image processing system 100 work together.

[0012] Figure 2 This is a schematic diagram illustrating an image processing method according to some embodiments of the present invention. Figure 7 This is a flowchart illustrating an image processing method according to some embodiments of the present invention. Please also refer to... Figure 1 , Figure 2 as well as Figure 7In some embodiments of the present invention, the image processing method includes steps S701 to S703 performed by the processing module 101. In step S701, the processing module 101 receives an input image 103 and obtains a front masking image 201 corresponding to the input image 103. The processing module 101 temporarily stores the input image 103 and the front masking image 201 in a frame buffer 200. The processing module 101 then enlarges the front masking image 201 to obtain a masking image 202, wherein the size of the masking image 202 is the same as the size of the input image 103. The pixel values ​​of the pixels in the masking image 202 represent the local information of each pixel in the input image 103.

[0013] In some embodiments of the present invention, the masking image 202 is an 8-bit image (i.e., the pixel values ​​of the image are stored in 8-bit units). The pixel values ​​of the pixels in the masking image 202 represent the relative inverse depth of each pixel in the input image 103. In other words, the larger the pixel value of the pixels in the masking image 202, the shallower the depth (i.e., the closer to the estimated distance from the lens). In this embodiment, the pixel values ​​of the pixels in the masking image 202 can distinguish the input image 103 into foreground and background. In some embodiments of the present invention, the masking image 202 is an 8-bit image, and the pixel values ​​of the pixels in the masking image 202 represent the classification label of each pixel in the input image 103 to which object it belongs. In this embodiment, the pixel values ​​of the pixels in the masking image 202 can distinguish the input image 103 into different objects. The masking image 202 will be further described in the following embodiments.

[0014] In step S702, processing module 101 performs an image quality function on input image 103 via image quality module 102 to obtain image quality image 203. In step S703, processing module 101 performs a pixel-weighted averaging procedure on input image 103 and image quality image 203 based on the value of each pixel in masked image 202 to obtain output image 204.

[0015] In the foregoing embodiments, by obtaining local information of the input image 103 through the masked image 202, and then using this local information to perform a pixel-weighted averaging process on the input image 103 and the image quality image 203, a specific image quality function can be performed on a local area of ​​the input image 103 without modifying the original architecture (e.g., circuit architecture) of the image quality module 102. When the image quality module 102 is implemented in hardware, no additional circuit area is required by not modifying its original architecture.

[0016] Taking the embodiment where the pixel value of the aforementioned masked image 202 represents the relative inverse depth of each pixel in the input image 103 as an example, since a larger pixel value of the masked image 202 indicates a shallower depth (that is, a closer distance to the estimated distance from the lens), if the desired effect is to obtain a stronger image quality effect in the foreground and a weaker image quality effect in the background, then during the pixel weighted averaging process, when the pixel value of the masked image 202 is larger, a larger weighting value is given to the corresponding pixel of the image quality image 203, and a smaller weighting value is given to the corresponding pixel of the input image 103; and when the pixel value of the masked image 202 is smaller, a smaller weighting value is given to the corresponding pixel of the image quality image 203, and a larger weighting value is given to the corresponding pixel of the input image 103.

[0017] In some embodiments of the present invention, the processing module 101 uses a bicubic interpolation algorithm to magnify the unmasked image 201 to obtain the masked image 202. In some embodiments of the present invention, the processing module 101 uses a nearest neighbor interpolation algorithm to magnify the unmasked image 201.

[0018] Figure 3 This is a block diagram of an image processing system according to an embodiment of the present invention. Figure 8 This is a flowchart illustrating an image processing method according to some embodiments of the present invention. Please also refer to... Figure 1 , Figure 3 as well as Figure 8 In some embodiments of the present invention, compared to image processing system 100, image processing system 300 further includes artificial intelligence module 301. In this embodiment, the aforementioned step S701 includes step S801. In step S801, processing module 101 obtains the front-masked image 201 of input image 103 via artificial intelligence module 301.

[0019] In some embodiments of the present invention, the artificial intelligence module 301 includes a depth message model. In step S801, the processing module 101 obtains the front masking image 201 corresponding to the input image 103 based on the depth message model.

[0020] In some embodiments of the present invention, the aforementioned depth message model is a MiDaS model. The MiDaS model can calculate relative inverse depth based on a single image; that is, the larger the pixel value of the depth estimation image output by the MiDaS model, the shallower the depth (i.e., the closer to the estimated distance from the lens). In this embodiment, the processing module 101 first reduces the input image 103 to the input size of the MiDaS model, and then inputs the reduced input image 103 into the MiDaS model. The processing module 101 uses the output of the MiDaS model as the front masking image 201 corresponding to the input image 103. It is worth noting that the depth message model can also be other monocular depth estimation models. Examples of such monocular depth estimation models include DORN (Deep Ordinal Regression Network), DenseDepth, DPT (Dense PredictionTransformers), GLPN (Global-Local Path Networks), or Marigold; the present invention does not limit this to any particular model. Among them, DORN and DenseDepth are models built using convolutional neural networks, DPT and GLPN are transformer-based models, and Marigold is a diffusion-based model.

[0021] Figure 9 This is a flowchart illustrating the front shielding image setting process according to some embodiments of the present invention. Please also refer to... Figure 1 , Figure 3 as well as Figure 9In some embodiments of the present invention, the artificial intelligence module 301 includes an image segmentation model. In step S801, the processing module 101 executes steps S901 to S904 to obtain a front masking image 201 corresponding to the input image 103. In step S901, the processing module 101 obtains an intermediate masking image corresponding to the input image 103 based on the image segmentation model. Each pixel in the intermediate masking image is assigned a classification label to indicate the object to which the pixel belongs. In step S902, for the current pixel of the front masking image 201, the processing module 101 sets the pixel value of the current pixel based on the classification label of the corresponding pixel in the intermediate masking image. In step S903, the processing module 101 determines whether the front masking image 201 has unprocessed pixels. If yes, it executes step S904; otherwise, it terminates the process. In step S904, in response to the front masking image 201 having unprocessed pixels, the processing module 101 sets the unprocessed pixel as the current pixel and returns to step S902.

[0022] In some embodiments of the present invention, the aforementioned image segmentation model is a U-Net model. In this embodiment, in step S901, the processing module 101 first reduces the input image 103 to the input size of the U-Net model, and then inputs the reduced input image 103 into the U-Net model. The processing module 101 uses the output of the U-Net model as the intermediate masking image corresponding to the input image 103. Each pixel in the intermediate masking image is assigned a classification label to indicate the object to which the pixel belongs. Next, in steps S902 and S903, the processing module 101 obtains the front masking image 201 based on the intermediate masking image.

[0023] In this embodiment, each object that the image segmentation model can identify has a preset value stored. For example, the preset value for a car is 0.5, and the preset value for the ground is 0.7. In step S902, for the current pixel of the front shielding image 201, the processing module 101 sets the pixel value of the current pixel to the default value corresponding to the object indicated by the classification label of the corresponding pixel in the intermediate shielding image. The corresponding pixel in the intermediate shielding image is the pixel at the same position as the current pixel in the intermediate shielding image.

[0024] For example, the current pixel in the front masking image 201 is the pixel at position (0, 0). Therefore, the corresponding pixel in the intermediate masking image is the pixel at position (0, 0) in the intermediate masking image. If the classification label of the corresponding pixel in the intermediate masking image indicates a vehicle, then the processing module 101 sets the pixel value of the current pixel (the pixel at position (0, 0) in the front masking image 201) to 0.5.

[0025] It is worth noting that the image segmentation model can also be FCN, DeconvNet, SegNet, DeepLab, RefineNet, PSPNet, or GSCNN, and this invention does not limit it.

[0026] Figure 10 as well as Figure 11 This is a flowchart illustrating an image processing method according to some embodiments of the present invention. Please also refer to... Figure 2 , Figure 7 , Figures 10-11 In some embodiments of the present invention, the aforementioned step S703 includes steps S1001 to S1004. In step S1001, the processing module 101 normalizes the masking image 202 so that the value of each pixel of the masking image 202 falls between 0 and 1, that is, falls within a closed interval. In [0, 1]. In step S1002, processing module 101 performs steps S1101 to S1103 on the current pixel in image quality image 102. In step S1003, processing module 101 determines whether image quality image 203 contains unprocessed pixels. If yes, it executes step S1004; otherwise, it terminates the process. In step S1004, in response to the presence of unprocessed pixels in image quality image 203, the unprocessed pixel is set as the current pixel, and the process returns to step S1002.

[0027] Figure 4A This is a schematic flowchart illustrating a pixel-weighted averaging process according to some embodiments of the present invention. Please also refer to... Figure 2 , Figure 4A , Figure 7 , Figures 10-11In step S1101, the processing module 101 multiplies the pixel value of the current pixel in the image quality image 203 by the pixel value of the corresponding current masking image pixel in the masking image 202 to obtain a first product. For example, the current pixel in the image quality image 203 is pixel 2031, and the position of pixel 2031 in the image quality image 203 is (0, 0). The pixel with the position (0, 0) in the masking image 202 is pixel 2021, therefore, the corresponding current masking image pixel in the masking image 202 is pixel 2021. Since the pixel value of pixel 2031 is v 00 The pixel value of pixel 2021 is a 00 Therefore, the first product is a. 00 ·v 00 .

[0028] In step S1102, processing module 101 subtracts the pixel value of the current masked image pixel from 1 to obtain a difference, and multiplies the difference by the pixel value of the input image pixel corresponding to the current pixel in the input image 103 to obtain a second product. Continuing the example above, the current masked image pixel is pixel 2021, therefore the difference is 1-a. 00 Since the current pixel in image quality image 203 is pixel 2031, and the position of pixel 2031 in image quality image 203 is (0, 0), and the pixel with position (0, 0) in input image 103 is pixel 1031, the input image pixel corresponding to the current pixel in input image 103 is pixel 1031. Since the pixel value of pixel 1031 is w... 00 Therefore, the second product is (1-a) 00 )·w 00 .

[0029] In step S1103, processing module 101 sets the pixel value of the output image pixel corresponding to the current pixel in the output image 204 to be the sum of the first product and the second product. Continuing the example above, the current pixel in the image quality image 203 is pixel 2031, and pixel 2031's position on the image quality image 203 is (0, 0). The pixel with position (0, 0) on the output image 204 is pixel 2041; therefore, the output image pixel corresponding to the current pixel in the output image 204 is pixel 2041. Processing module 101 sets the pixel value of pixel 2041 to a. 00 ·v 00 +(1-a 00 )·w 00 .

[0030] For another example, the current pixel in image quality image 203 is pixel 2032, and the position of pixel 2032 in image quality image 203 is (i, j), where i and j are both non-negative integers. The pixel at position (i, j) in masking image 202 is pixel 2022. Therefore, the current masking image pixel corresponding to the current pixel in masking image 202 is pixel 2022. Since the pixel value of pixel 2032 is v... ij The pixel value of pixel 2022 is a ij Therefore, the first product is a. ij ·v ij The current masked image has 2022 pixels, therefore the difference is 1-a. ij The pixel at position (i, j) in input image 103 is pixel 1032. Therefore, the input image pixel corresponding to the current pixel in input image 103 is pixel 1032. Since the pixel value of pixel 1032 is w ij Therefore, the second product is (1-a) ij )·w ij The pixel at position (i, j) on the output image 204 is pixel 2042. Therefore, the output image pixel corresponding to the current pixel in the output image 204 is pixel 2042. The processing module 101 sets the pixel value of pixel 2042 to a. ij ·v ij +(1-a ij )·w ij .

[0031] In this embodiment, the aforementioned pixel-weighted averaging procedure includes steps S1002 to S1004.

[0032] Following the aforementioned embodiment where the masked image 202 is an 8-bit image and the pixel values ​​of the masked image 202 represent the relative inverse depth of each pixel in the input image 103, in step S1001, the processing module 101 obtains the maximum pixel value (max) and the minimum pixel value (min) in the masked image 202. The processing module 101 subtracts min from the pixel value of each pixel in the masked image 202 and multiplies it by 1 / (max-min) to obtain the normalized masked image 202, where the value of each pixel in the normalized masked image 202 falls between 0 and 1. Since the closer the pixel value of the masked image 202 is to 1, the shallower the depth (i.e., the closer to the estimated distance of the lens), after the processing module 101 completes the process... Figure 10 After the illustrated process, the output image 204 has a stronger image quality effect in the foreground and a weaker image quality effect in the background compared to the input image 103.

[0033] Figure 4BThis is a schematic diagram of the difference image according to some embodiments of the present invention. The processing module 101 can obtain the same result as in steps S1002 to S1004 through point-by-point matrix multiplication and addition operations. Please also refer to... Figure 2 , Figure 4A , Figure 4B , Figure 7 , Figures 10-11 In some embodiments of the present invention, the processing module 101 subtracts the masking image 202 from a full tensor (i.e., tensor values ​​are all 1) to obtain the difference image 401. For example, the pixel value of the aforementioned pixel 2021 is a. 00 The pixel value of pixel 2022 is a ij In this embodiment, the pixel at position (0, 0) on the difference image 401 is pixel 4011, and the pixel at position (i, j) on the difference image 401 is pixel 4012. The pixel value of pixel 4011 is 1-a. 00 The pixel value of pixel 4012 is 1-a ij .

[0034] Processing module 101 performs point-by-point multiplication on the masked image 202 and the image quality image 203 to obtain a first product image. Processing module 101 performs point-by-point multiplication on the difference image 401 and the input image 103 to obtain a second product image. Processing module 101 performs matrix addition on the first product image and the second product image to obtain an output image 204. That is, if I represents a full sheet, the relationship between the output image 204, the masked image 202, the image quality image 203, and the input image 103 is: Output image 204 = Masked image 202 ⊙ Image quality image 203 + Difference image 401 ⊙ Input image 103, where ⊙ represents point-by-point multiplication.

[0035] It is worth noting that when the input image 103 is a color image, the image quality image 203 is also a color image. The processing module 101 performs point-by-point multiplication on the masked image 202 and the image quality image 203 (i.e., calculates: masked image 202 ⊙ image quality image 203), indicating that the processing module 101 performs point-by-point multiplication on the R, G, and B components of the image quality image 203 using the masked image 202. The processing module 101 performs point-by-point multiplication on the difference image 401 and the input image 103 (i.e., calculates: difference image 401 ⊙ input image 103), indicating that the processing module 101 performs point-by-point multiplication on the R, G, and B components of the input image 103 using the difference image 401.

[0036] Figure 5 This is a schematic diagram illustrating an image processing method according to some embodiments of the present invention. Figure 12 This is a flowchart illustrating an image processing method according to some embodiments of the present invention. Please also refer to... Figure 1 , Figure 5 as well as Figure 12 In some embodiments of the present invention, the image processing method includes performing step S1201 after step S703. In step S1201, an image processing procedure is performed on the output image 204 to obtain a processed output image 502. In some embodiments of the present invention, the image processing procedure includes: processing module 101 using a bilateral filter to filter the output image 204 to make the boundaries between regions in the output image 204 appear more natural.

[0037] If it is necessary to perform the same or multiple image quality functions on a local area in the output image 204, the processing module 101 can obtain multiple pre-masking images, and then perform a pixel-weighted averaging procedure on the input image 103 in stages using the information from each pre-masking image. Please refer to [further details omitted]. Figure 5 ,exist Figure 5 In the illustrated embodiment, after obtaining the processed output image 502, the processing module 101 obtains the pre-masking image 501 and magnifies it to obtain a masking image 505. Then, the processing module 101 performs an image quality function on the processed output image 502 via the image quality module 503 to obtain an image quality image 504. The processing module 101 then performs a pixel-weighted averaging procedure on the processed output image 502 and the image quality image 504 based on the value of each pixel in the masking image 505 to obtain an output image 506. The processing module 101 performs an image processing procedure on the output image 506 to obtain a processed output image 507.

[0038] Figure 6 These are block diagrams of electronic device systems illustrated according to some embodiments of the present invention. Figure 6 As shown, at the hardware level, electronic device 600 includes a processing unit 601, internal memory 602, and non-volatile memory 603. Internal memory 602 is, for example, random-access memory (RAM). Non-volatile memory 603 is, for example, at least one disk storage device. Of course, electronic device 600 may also include other hardware required for its functions.

[0039] Internal memory 602 and non-volatile memory 603 are used to store programs, which may include program code and computer operation instructions. Internal memory 602 and non-volatile memory 603 provide instructions and data to processing unit 601. Processing unit 601 reads the corresponding computer program from non-volatile memory 603 into internal memory 602 and then runs it.

[0040] The processing unit 601 may be an integrated circuit chip with signal processing capabilities. In implementation, the methods and steps disclosed in the foregoing embodiments can be completed by integrated logic circuits in the hardware or by instructions in software within the processing unit 601. The processing unit 601 may be a general-purpose processor, including a central processing unit, a digital signal processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, capable of implementing or executing the methods and steps disclosed in the foregoing embodiments.

[0041] This specification also provides a non-transitory computer-readable storage medium that stores at least one instruction. When executed by the processing unit 601 of the electronic device 600, the at least one instruction enables the processing unit 601 of the electronic device 600 to perform the methods and steps disclosed in the foregoing embodiments.

[0042] Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other internal memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, non-transitory computer-readable media does not include transient media, such as modulated data signals and carrier waves.

[0043] The foregoing embodiments provide an image processing system, image processing method, a non-transitory computer-readable recording medium with memory program, and a non-transitory computer program product. By masking an image to obtain local information of an input image, and then using this local information to perform a pixel-weighted averaging process on the input image and the image quality image, the system can achieve the effect of performing specific image quality functions on a local area of ​​the input image without modifying the original architecture (e.g., circuit architecture) of the image quality module. When the image quality module is implemented in hardware, no additional circuit area is required by not modifying the original architecture of the image quality module.

[0044] Although the present invention has been disclosed above by way of embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make some modifications and refinements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims. [Symbol Explanation]

[0045] 100, 300: Image processing system 101: Processing Module 102, 503: Image quality module 103: Input Image 201, 501: Anterior shielding image 200: Frame buffer 202, 505: Blocked Image 203, 504: Image quality images 204: Output Image 301: Artificial Intelligence Module 401: Poor Image 1031, 1032, 2021, 2022, 2031, 2032, 2041, 2042, 4011, 4012: pixels 506: Output Image 502, 507: Processed output images 600: Electronic Equipment 601: Processing Unit 602: Internal Memory 603: Non-volatile memory S701~S703, S801, S901~S904, S1001~S1004, S1101~S1103, S1201: Steps

Claims

1. An image processing system, characterized in that, Include: An image quality module, configured to perform an image quality function; and A processing module is configured to receive an input image and perform the following steps: (a) Obtain a front masking image of the input image and enlarge the front masking image to obtain a masking image, wherein a size of the masking image is the same as a size of the input image; (b) Performing the image quality function on the input image via the image quality module to obtain an image quality image; and (c) Based on a value of each of the multiple pixels of the masked image, perform a pixel-weighted averaging procedure on the input image and the image quality image to obtain an output image.

2. The image processing system as described in claim 1, characterized in that, The image processing system includes an artificial intelligence module, which obtains the front-masked image of the input image via the artificial intelligence module.

3. The image processing system as described in claim 2, characterized in that, The artificial intelligence module includes a deep message model, and step (a) includes: obtaining the front masking image corresponding to the input image based on the deep message model.

4. The image processing system as described in claim 2, characterized in that, The artificial intelligence module includes an image segmentation model, and step (a) includes the following steps to obtain the front masking image: (a1) obtaining an intermediate masking image corresponding to the input image based on the image segmentation model; (a2) for a current pixel in the front masking image, setting a pixel value of the current pixel based on a classification label of a corresponding pixel in the intermediate masking image. And (a3) ​​in response to the fact that the front masking image has an unprocessed pixel, set the unprocessed pixel as the current pixel and return to step (a2).

5. The image processing system as described in claim 1, characterized in that, Step (c) includes: (c 1) Regularize the shielded image; (c 2) Perform the following on a current pixel in the image quality image: (c21) Multiply the pixel value of the current pixel by the pixel value of a current masking image pixel in the masking image that corresponds to the current pixel to obtain a first product; (c22) Subtract the pixel value of the current masked image pixel from 1 to obtain a difference; and multiply the difference by a pixel value of an input image pixel corresponding to the current pixel in the input image to obtain a second product; (c23) Set a pixel value in the output image corresponding to the current pixel to the sum of the first product and the second product; and (c 3) In response to the presence of an unprocessed pixel in the image quality image, set the unprocessed pixel as the current pixel and return to step (c 2).

6. The image processing system as described in claim 1, characterized in that, The processing module is configured to perform an image processing procedure on the output image after step (c) to obtain a processed output image.

7. The image processing system as described in claim 6, characterized in that, The image processing program includes: performing a filtering process on the output image using a bilateral filter.

8. An image processing method, executed by a processing module, characterized in that, This image processing method includes: (a) Receive an input image; obtain a front masking image of the input image and enlarge the front masking image to obtain a masking image, wherein a size of the masking image is the same as a size of the input image; (b) Performing an image quality function on the input image via an image quality module to obtain an image quality image; and (c) Based on a value of each of the multiple pixels of the masked image, perform a pixel-weighted averaging procedure on the input image and the image quality image to obtain an output image.

9. The image processing method as described in claim 8, characterized in that, Step (a) includes: obtaining the front-masked image of the input image via an artificial intelligence module.

10. A non-transitory computer program product, storing at least one instruction, characterized in that, When the at least one instruction is executed by a processing unit, the processing unit is caused to perform the method as described in any one of claims 8 to 9.

Images