Digital population reasoning image restoration method and system

Abstract

The invention provides a digital population reasoning image restoration method and system, and the method comprises the steps: obtaining an original template video, and carrying out the digital population reasoning of the original template video through a reasoning tool, so as to obtain a reasoning video; preprocessing the original template video and the reasoning video to obtain an original video frame and a reasoning video frame; performing preset super-resolution processing on the reasoning video frame to obtain a super-resolution image frame; extracting a first face mark point of the original video frame and a second face mark point of the super-division image frame by adopting a preset recognition algorithm, and adjusting the second face mark point of the super-division image frame based on the first face mark point to obtain an adjusted image frame; the mask image of the target area is determined based on the adjusted image frame, the repaired image is determined based on the mask image, the adjusted image frame and the original video frame, and the clear repaired image without chin problems can be obtained by determining the mask image and overlapping the adjusted image frame and the original video frame.

Description

Technical Field

[0001] The present invention belongs to the technical field of image restoration, and in particular relates to a digital population-based inference image restoration method and system. Background Art

[0002] Digital population shape reasoning is one of the necessary technologies to realize the broadcasting digital human. In order to complete the reasoning quickly, it is usually necessary to cut out the rectangular position from the nostrils to the chin through face recognition, and then perform lip shape reasoning. Before the reasoning, the digital human is closed-mouthed, and after the reasoning, it is the process of speaking. In this case, the cut rectangular frame may cause the chin to protrude outside the rectangular frame due to the excessive amplitude of speaking, resulting in a missing chin.

[0003] Due to the limitations of hardware and computing power, digital population inference will appear partially blurred and the image will be color-shifted after inference, resulting in an obvious dividing line at the junction of the cut rectangular blocks with the original image when they are pasted back;

[0004] In summary, the images or videos after digital population inference in the prior art have the problems of color cast, local blur and missing chin. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a digital population-based inference image restoration method and system, which are used to solve the technical problems raised in the prior art, so as to reduce the user's time cost and ensure that the user can find the target terminal in the shortest time.

[0006] In a first aspect, an embodiment of the present invention provides the following technical solution, a method for restoring a digital population-based inference image, comprising:

[0007] Obtaining an original template video, and using an inference tool to perform digital population type inference on the original template video to obtain an inference video;

[0008] Preprocessing the original template video and the inference video to obtain an original video frame and an inference video frame;

[0009] Performing a preset super-resolution process on the inference video frame to obtain a super-resolution image frame;

[0010] Extracting a first face marker of the original video frame and a second face marker of the super-resolution image frame using a preset recognition algorithm, and adjusting the second face marker of the super-resolution image frame based on the first face marker to obtain an adjusted image frame;

[0011] A mask image of a target area is determined based on the adjusted image frame, and a repaired image is determined based on the mask image, the adjusted image frame and the original video frame.

[0012] Compared with the prior art, the beneficial effects of the present application are as follows: the present invention first obtains the original template video, and uses the inference tool to perform digital population type inference on the original template video to obtain the inference video; then the original template video and the inference video are preprocessed to obtain the original video frame and the inference video frame; then the inference video frame is subjected to preset super-resolution processing to obtain the super-resolution image frame; then a preset recognition algorithm is used to extract the first face marker point of the original video frame and the second face marker point of the super-resolution image frame, and the second face marker point of the super-resolution image frame is adjusted based on the first face marker point to obtain the adjusted image frame; then a mask image of the target area is determined based on the adjusted image frame, and a repaired image is determined based on the mask image, the adjusted image frame and the original video frame. The present invention can obtain a clear digital human face image with no chin problems, that is, a repaired image, by determining the mask image and superimposing the adjusted image frame with the original video frame.

[0013] Preferably, the reasoning tool is a wav2lip open source library.

[0014] Preferably, the step of preprocessing the original template video and the inference video to obtain the original video frame and the inference video frame is specifically as follows:

[0015] The original template video and the inference video are cropped, rotated, normalized, smoothed and denoised, and split frame by frame in sequence to obtain original video frames and inference video frames.

[0016] Preferably, in the step of performing preset super-resolution processing on the inference video frame to obtain a super-resolution image frame, the preset super-resolution processing is specifically GFPGAN equal resolution super-resolution processing.

[0017] Preferably, the preset recognition algorithm is specifically the shape_predictor_68_face_landmarks.dat algorithm.

[0018] Preferably, the step of adjusting the second face marker points of the super-resolution image frame based on the first face marker points to obtain an adjusted image frame comprises:

[0019] Calculate the center point V of all the first face landmarks c :

[0020]

[0021] In the formula, represents the i-th first face landmark point, and N represents the number of first face landmark points;

[0022] Based on the center point V c Determine the first vector With the second vector

[0023]

[0024] Where V cf 、V o Respectively represent the set of all second face landmarks and the set of all first face landmarks;

[0025] Based on the first vector With the second vector Determine the elongation rate:

[0026]

[0027] The second facial landmark point is adjusted based on the elongation rate Rate to obtain an adjustment point Determine the adjustment image frame based on the adjustment point:

[0028]

[0029] Preferably, the step of determining a mask image of a target area based on the adjusted image frame, and determining a repaired image based on the mask image, the adjusted image frame and the original video frame comprises:

[0030] Extracting a target image of a target area of ​​the adjusted image frame, wherein the target area includes a mouth area and a chin area;

[0031] Performing Gaussian blur, polygon filling and binarization processing on the target image to obtain a mask image;

[0032] The adjusted image frame is superimposed with the original video frame using the mask image as a transparent channel to obtain a repaired image Result:

[0033] Result = img cf ×mask+img original ×(1―mask);

[0034] In the formula, img cf 、img original They represent the adjusted image frame and the original video frame respectively, and mask represents the mask image.

[0035] In a second aspect, an embodiment of the present invention provides the following technical solution, a digital population-based inference image restoration system, the system comprising:

[0036] An inference module is used to obtain an original template video and use an inference tool to perform digital population type inference on the original template video to obtain an inference video;

[0037] A processing module, used for preprocessing the original template video and the inference video to obtain an original video frame and an inference video frame;

[0038] A super-resolution module, used for performing a preset super-resolution process on the inference video frame to obtain a super-resolution image frame;

[0039] an adjustment module, configured to extract a first face marker of the original video frame and a second face marker of the super-resolution image frame by using a preset recognition algorithm, and adjust the second face marker of the super-resolution image frame based on the first face marker to obtain an adjusted image frame;

[0040] The restoration module is used to determine a mask image of a target area based on the adjusted image frame, and to determine a restoration image based on the mask image, the adjusted image frame and the original video frame.

[0041] In a third aspect, an embodiment of the present invention provides the following technical solution: a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned digital population-type inference image restoration method when executing the computer program.

[0042] In a fourth aspect, an embodiment of the present invention provides the following technical solution: a computer-readable storage medium having a computer program stored thereon, and the computer program, when executed by a processor, implements the above-mentioned digital population-type inference image restoration method. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0044] Figure 1 A flowchart of a digital population-based inference image restoration method provided by an embodiment of the present invention;

[0045] Figure 2 A structural block diagram of a digital population-based inference image restoration system provided by an embodiment of the present invention;

[0046] Figure 3 A schematic diagram of the hardware structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the embodiments of the present invention, and should not be construed as limiting the present invention.

[0048] Embodiment 1

[0049] In the first embodiment of the present invention, Figure 1 As shown, a digital population-based inference image restoration method comprises:

[0050] S1. Obtain an original template video, and use an inference tool to perform digital population type inference on the original template video to obtain an inference video;

[0051] Specifically, in this application, digital population type inference is performed on the original template video through the wav2lip open source library to obtain an inference video.

[0052] S2, preprocessing the original template video and the inference video to obtain an original video frame and an inference video frame;

[0053] Specifically, the step S2 is as follows:

[0054] The original template video and the inference video are cropped, rotated, normalized, smoothed and denoised, and split frame by frame in sequence to obtain original video frames and inference video frames.

[0055] S3, performing a preset super-resolution process on the inference video frame to obtain a super-resolution image frame;

[0056] Specifically, in step S3, the preset super-resolution processing is specifically GFPGAN equal resolution super-resolution processing, and the clarity of the image can be further improved through the preset super-resolution processing.

[0057] S4, extracting a first face marker point of the original video frame and a second face marker point of the super-resolution image frame by using a preset recognition algorithm, and adjusting the second face marker point of the super-resolution image frame based on the first face marker point to obtain an adjusted image frame;

[0058] Wherein, in step S4, the preset recognition algorithm is specifically the shape_predictor_68_face_landmarks.dat algorithm, which is specifically a face landmark point recognition algorithm based on a face recognition 68 feature point detection database, and can effectively identify the landmark points of the face in the image;

[0059] Moreover, in the specific recognition process, the marker points identified by the preset recognition algorithm have different numbers, and the marker points [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14] are selected as the first face marker points or the second face marker points. After that, the convex hull calculation is performed to form the target image of the target area in the subsequent steps.

[0060] Wherein, the step S4 comprises:

[0061] S41, calculating the center point V of all the first facial landmarks c :

[0062]

[0063] In the formula, represents the i-th first face landmark point, and N represents the number of first face landmark points.

[0064] S42, based on the center point V c Determine the first vector With the second vector

[0065]

[0066] Where V cf 、V o They respectively represent the set of all second face landmarks and the set of all first face landmarks.

[0067] S43, based on the first vector With the second vector Determine the elongation rate:

[0068]

[0069] S44, adjusting the second facial landmark point based on the elongation rate Rate to obtain an adjustment point Determine the adjustment image frame based on the adjustment point:

[0070]

[0071] Specifically, in actual situations, the super-resolution image frame and the original video frame will have inconsistent edges, so edge contraction or expansion is required, that is, all the first face marker points are equal to the center point, and then the center point is subtracted from all the marker points in the super-resolution image frame and the original video frame to obtain two vectors. The elongation rate can be obtained based on the two vectors. When the elongation rate is less than 0, it indicates contraction, and when it is greater than 0, it indicates expansion. In this way, the adjustment points can be adjusted accordingly according to the edges of the original video frame to obtain the adjusted image frame.

[0072] S5. Determine a mask image of the target area based on the adjusted image frame, and determine a restored image based on the mask image, the adjusted image frame and the original video frame.

[0073] Wherein, the step S5 comprises:

[0074] S51, extracting a target image of a target area of ​​the adjusted image frame, wherein the target area includes a mouth area and a chin area.

[0075] S52, performing Gaussian blur, polygon filling and binarization processing on the target image to obtain a mask image.

[0076] S53, superimposing the adjusted image frame and the original video frame with the mask image as a transparent channel to obtain a repaired image Result:

[0077] Result = img cf ×mask+img original ×(1―mask);

[0078] In the formula, img cf 、img original They represent the adjusted image frame and the original video frame respectively, and mask represents the mask image.

[0079] Specifically, the mask image serves as a transparent channel. For the original video frame, it is black and transparent while white is transparent. For the adjusted image frame, it is black and transparent while white is not transparent. Therefore, two partially transparent images can be obtained. The two images are then superimposed to obtain a repaired image, and the repaired image is a digital human face image with clear synthesis and no problems with the chin position.

[0080] The digital population type inference image restoration method provided in the first embodiment of the present invention first obtains the original template video, uses the inference tool to perform digital population type inference on the original template video to obtain the inference video; then pre-processes the original template video and the inference video to obtain the original video frame and the inference video frame; then performs preset super-resolution processing on the inference video frame to obtain the super-resolution image frame; then uses the preset recognition algorithm to extract the first face marker point of the original video frame and the second face marker point of the super-resolution image frame, and adjusts the second face marker point of the super-resolution image frame based on the first face marker point to obtain the adjusted image frame; then determines the mask image of the target area based on the adjusted image frame, and determines the restoration image based on the mask image, the adjusted image frame and the original video frame. The present invention can obtain a clear digital human face image with no chin problem, that is, a restoration image, by determining the mask image and superimposing the adjusted image frame with the original video frame.

[0081] Embodiment 2

[0082] like Figure 2 As shown, in the second embodiment of the present invention, a digital population type inference image restoration system is provided, and the system includes:

[0083] Inference module 1, used to obtain an original template video, and use an inference tool to perform digital population type inference on the original template video to obtain an inference video;

[0084] Processing module 2, used for preprocessing the original template video and the inference video to obtain an original video frame and an inference video frame;

[0085] The super-resolution module 3 is used to perform a preset super-resolution process on the inference video frame to obtain a super-resolution image frame;

[0086] An adjustment module 4 is used to extract a first face marker point of the original video frame and a second face marker point of the super-resolution image frame by using a preset recognition algorithm, and adjust the second face marker point of the super-resolution image frame based on the first face marker point to obtain an adjusted image frame;

[0087] A restoration module 5, configured to determine a mask image of a target area based on the adjusted image frame, and to determine a restoration image based on the mask image, the adjusted image frame and the original video frame;

[0088] The processing module 2 is specifically used for:

[0089] The original template video and the inference video are cropped, rotated, normalized, smoothed and denoised, and split frame by frame in sequence to obtain original video frames and inference video frames.

[0090] The adjustment module 4 comprises:

[0091] The first calculation submodule is used to calculate the center point V of all the first facial landmarks. c :

[0092]

[0093] In the formula, represents the i-th first face landmark point, and N represents the number of first face landmark points;

[0094] The second calculation submodule is used to calculate the value of the central point V c Determine the first vector With the second vector

[0095]

[0096] Where V cf 、V o Respectively represent the set of all second face landmarks and the set of all first face landmarks;

[0097] The third calculation submodule is used to calculate the first vector With the second vector Determine the elongation rate:

[0098]

[0099] The fourth calculation submodule is used to adjust the second facial landmark point based on the extension rate Rate to obtain an adjustment point Determine the adjustment image frame based on the adjustment point:

[0100]

[0101] The repair module 5 comprises:

[0102] An extraction submodule, used for extracting a target image of a target area of ​​the adjusted image frame, wherein the target area includes a mouth area and a chin area;

[0103] A processing submodule, used for performing Gaussian blur, polygon filling and binarization processing on the target image to obtain a mask image;

[0104] The superposition submodule is used to superimpose the adjusted image frame with the original video frame using the mask image as a transparent channel to obtain a repaired image Result:

[0105] Result = img cf ×mask+img original ×(1―mask);

[0106] In the formula, img cf 、img original They represent the adjusted image frame and the original video frame respectively, and mask represents the mask image.

[0107] In some other embodiments of the present invention, the embodiments of the present invention provide the following technical solution: a computer, comprising a memory 102, a processor 101, and a computer program stored in the memory 102 and executable on the processor 101, wherein the processor 101 implements the digital population-type reasoning image restoration method as described above when executing the computer program.

[0108] Specifically, the processor 101 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiment of the present invention.

[0109] Among them, the memory 102 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 102 may include a hard disk drive (HDD), a floppy disk drive, a solid state drive (SSD), a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 102 may include a removable or non-removable (or fixed) medium. Where appropriate, the memory 102 may be inside or outside a data processing device. In a specific embodiment, the memory 102 is a non-volatile memory. In a specific embodiment, the memory 102 includes a read-only memory (ROM) and a random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (Programmable Read-Only Memory, PROM for short), an erasable PROM (Erasable Programmable Read-Only Memory, EPROM for short), an electrically erasable PROM (Electrically Erasable Programmable Read-Only Memory, EEPROM for short), an electrically alterable ROM (Electrically Alterable Read-Only Memory, EAROM for short) or a flash memory (FLASH) or a combination of two or more of these. Under appropriate circumstances, the RAM can be a static random access memory (SRAM) or a dynamic random access memory (DRAM), wherein the DRAM can be a fast page mode dynamic random access memory (FPMDRAM), an extended data output dynamic random access memory (EDODRAM), a synchronous dynamic random access memory (SDRAM), etc.

[0110] The memory 102 may be used to store or cache various data files that need to be processed and / or used for communication, as well as possible computer program instructions executed by the processor 101 .

[0111] The processor 101 implements the above-mentioned digital population-type inference image restoration method by reading and executing computer program instructions stored in the memory 102 .

[0112] In some embodiments, the computer may further include a communication interface 103 and a bus 100. Figure 3 As shown, the processor 101, the memory 102, and the communication interface 103 are connected via a bus 100 and communicate with each other.

[0113] The communication interface 103 is used to implement communication between the modules, devices, units and / or equipment in the embodiment of the present invention. The communication interface 103 can also implement data communication with other components such as: external devices, image / data acquisition equipment, databases, external storage, and image / data processing workstations.

[0114] The bus 100 includes hardware, software or both, and couples the components of the computer device to each other. The bus 100 includes but is not limited to at least one of the following: a data bus, an address bus, a control bus, an expansion bus, and a local bus. By way of example and not limitation, bus 100 may include Accelerated Graphics Port (AGP) or other graphics bus, Extended Industry Standard Architecture (EISA) bus, Front Side Bus (FSB), Hyper Transport (HT) interconnect, Industry Standard Architecture (ISA) bus, InfiniBand interconnect, Low Pin Count (LPC) bus, memory bus, Micro Channel Architecture (MCA) bus, Peripheral Component Interconnect (PCI) bus, PCI-Express (PCI-X) bus, Serial Advanced Technology Attachment (SATA) bus, Video Electronics Standards Association Local Bus (VLB) bus, or other suitable bus or a combination of two or more of these. Where appropriate, bus 100 may include one or more buses. Although embodiments of the present invention describe and illustrate a particular bus, the present invention contemplates any suitable bus or interconnect.

[0115] The computer can execute the digital population type reasoning image restoration method of the present invention based on the digital population type reasoning image restoration system, thereby realizing the digital population type reasoning image restoration.

[0116] In some further embodiments of the present invention, in combination with the above-mentioned digital population-type inference image restoration method, the embodiments of the present invention provide the following technical solutions: a storage medium having a computer program stored thereon, and the computer program implements the above-mentioned digital population-type inference image restoration method when executed by a processor.

[0117] Those skilled in the art will appreciate that the logic and / or steps represented in the flowchart or otherwise described herein, for example, may be considered as an ordered list of executable instructions for implementing logical functions, and may be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For purposes of this specification, a "computer-readable medium" may be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0118] More specific examples of readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0119] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or a combination thereof: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0120] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0121] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A digital population-based inference image restoration method, characterized in that: include: Obtaining an original template video, and using an inference tool to perform digital population type inference on the original template video to obtain an inference video; Preprocessing the original template video and the inference video to obtain an original video frame and an inference video frame; Performing a preset super-resolution process on the inference video frame to obtain a super-resolution image frame; Extracting a first face marker of the original video frame and a second face marker of the super-resolution image frame using a preset recognition algorithm, and adjusting the second face marker of the super-resolution image frame based on the first face marker to obtain an adjusted image frame; A mask image of a target area is determined based on the adjusted image frame, and a repaired image is determined based on the mask image, the adjusted image frame and the original video frame.

2. The digital population type inference image restoration method according to claim 1, characterized in that: The reasoning tool is the wav2lip open source library.

3. The digital population type inference image restoration method according to claim 1, characterized in that: The step of preprocessing the original template video and the inference video to obtain the original video frame and the inference video frame is specifically as follows: The original template video and the inference video are cropped, rotated, normalized, smoothed and denoised, and split frame by frame in sequence to obtain original video frames and inference video frames.

4. The digital population type inference image restoration method according to claim 1, characterized in that: In the step of performing a preset super-resolution process on the inference video frame to obtain a super-resolution image frame, the preset super-resolution process is specifically a GFPGAN equal resolution super-resolution process.

5. The digital population type inference image restoration method according to claim 1, characterized in that: The preset recognition algorithm is specifically the shape_predictor_68_face_landmarks.dat algorithm.

6. The digital population type inference image restoration method according to claim 1, characterized in that: The step of adjusting the second face marker points of the super-resolution image frame based on the first face marker points to obtain an adjusted image frame comprises: Calculate the center point V of all the first face landmarks c : In the formula, represents the i-th first face landmark point, and N represents the number of first face landmark points; Based on the center point V c Determine the first vector V cf1 With the second vector V o1 : V cf1 =V cf ―V c ;V o1 =V o ―V c ; Where V cf 、V o Respectively represent the set of all second face landmarks and the set of all first face landmarks; Based on the first vector V cf1 With the second vector V o1 Determine the elongation rate: Rate= ( |V c f1 | / V o1 ) ―1; The second facial landmark point is adjusted based on the elongation rate Rate to obtain an adjustment point V cf2 , determining the adjusted image frame based on the adjustment point: V cf2 =V c +V cf1 ×(Rate+1)。 7. The digital population type inference image restoration method according to claim 1, characterized in that: The steps of determining a mask image of a target area based on the adjusted image frame, and determining a repaired image based on the mask image, the adjusted image frame and the original video frame include: Extracting a target image of a target area of ​​the adjusted image frame, wherein the target area includes a mouth area and a chin area; Performing Gaussian blur, polygon filling and binarization processing on the target image to obtain a mask image; The adjusted image frame is superimposed with the original video frame using the mask image as a transparent channel to obtain a repaired image Result: Result=img cf ×mask+img original ×(1―mask); In the formula, img cf 、img original They represent the adjusted image frame and the original video frame respectively, and mask represents the mask image.

8. A digital population-based inference image restoration system, characterized in that: include: An inference module is used to obtain an original template video and use an inference tool to perform digital population type inference on the original template video to obtain an inference video; A processing module, used for preprocessing the original template video and the inference video to obtain an original video frame and an inference video frame; A super-resolution module, used for performing a preset super-resolution process on the inference video frame to obtain a super-resolution image frame; an adjustment module, configured to extract a first face marker of the original video frame and a second face marker of the super-resolution image frame by using a preset recognition algorithm, and adjust the second face marker of the super-resolution image frame based on the first face marker to obtain an adjusted image frame; The restoration module is used to determine a mask image of a target area based on the adjusted image frame, and to determine a restoration image based on the mask image, the adjusted image frame and the original video frame.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the digital population-type inference image restoration method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the digital population-type inference image restoration method according to any one of claims 1 to 7 is implemented.

Images