Image optimization method and device based on non-reference image, equipment and storage medium

Abstract

The invention discloses an image optimization method and device based on a non-reference image, equipment and a storage medium. The image optimization method based on the non-reference image comprises the steps of generating a to-be-optimized image set according to at least one piece of to-be-optimized image data; based on a preset image optimization algorithm, performing image optimization on the to-be-optimized image to generate a candidate image; generating a target image quality score of the candidate image based on at least one non-reference image quality evaluation index of the candidate image, and performing optimization updating on a candidate hyper-parameter combination based on the target image quality score; and performing image optimization on the to-be-optimized image again based on the optimized and updated target hyper-parameter combination to generate a target image. According to the technical scheme, the applicability and engineering feasibility of image optimization in a complex image scene are improved.

Description

Technical Field

[0001] This application relates to the field of image processing technology, specifically to an image optimization method, apparatus, device, and storage medium based on a no-reference image. Background Technology

[0002] In fields such as natural images, fluorescence images, and X-ray images, restoring images with noise, blurriness, or low resolution to high-quality, clear images through image restoration and other methods has become an important research direction in the field of image processing.

[0003] Image restoration typically employs iterative algorithms, whose performance is highly dependent on iterative hyperparameters (such as regularization weights, penalty factors, step size / relaxation, and number of iterations). Current hyperparameter selection methods mainly include: 1) Generalized cross-validation (GCV): typically used in linear inverse problems / ridge regression frameworks, it selects hyperparameters by minimizing the problem-related GCV function, requiring approximate linearity and Gaussian noise assumptions, making it difficult to adapt to nonlinear / non-convex iterative algorithms and complex priors; 3) Stern unbiased risk estimation (SURE): has high requirements for the differentiability of noise distribution and denoiser, but faces certain challenges in handling non-smooth regions and non-Gaussian noise. Summary of the Invention

[0004] This application provides an image optimization method, apparatus, device, and storage medium based on a no-reference image, to improve the robustness of image optimization and its applicability and engineering feasibility in complex image scenarios.

[0005] According to one aspect of this application, an image optimization method based on a no-reference image is provided, the method comprising:

[0006] A set of images to be optimized is generated based on at least one image data to be optimized; wherein, the image to be optimized refers to a degraded image;

[0007] Based on a preset image optimization algorithm, image optimization is performed on the image to be optimized to generate candidate images; wherein, the image optimization algorithm includes candidate hyperparameter combinations to be optimized;

[0008] Based on at least one no-reference image quality evaluation index of the candidate image, a target image quality score is generated for the candidate image, and the candidate hyperparameter combination is optimized and updated based on the target image quality score; wherein, the target image quality score is used to characterize the image quality of the candidate image under the hyperparameter conditions corresponding to the candidate hyperparameter combination;

[0009] Based on the optimized and updated combination of target hyperparameters, the image to be optimized is optimized again to generate the target image.

[0010] According to another aspect of this application, an image optimization apparatus based on a no-reference image is provided, the apparatus comprising:

[0011] An image set generation module is used to generate an image set to be optimized based on at least one image data to be optimized; wherein, the image to be optimized refers to a degraded image;

[0012] A candidate image generation module is used to optimize an image to be optimized based on a preset image optimization algorithm to generate candidate images; wherein, the image optimization algorithm includes candidate hyperparameter combinations to be optimized;

[0013] The hyperparameter optimization module is used to generate a target image quality score for the candidate image based on at least one no-reference image quality evaluation index, and to optimize and update the candidate hyperparameter combination based on the target image quality score; wherein, the target image quality score is used to characterize the image quality of the candidate image under the hyperparameter conditions corresponding to the candidate hyperparameter combination.

[0014] The target image generation module is used to perform image optimization on the image to be optimized again and generate the target image based on the optimized and updated target hyperparameter combination.

[0015] According to another aspect of this application, an electronic device is provided, the electronic device comprising:

[0016] One or more processors;

[0017] Memory, used to store one or more programs;

[0018] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the image optimization methods based on no-reference images provided in the embodiments of this application.

[0019] According to another aspect of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements any of the image optimization methods based on a no-reference image provided in the embodiments of this application.

[0020] According to another aspect of this application, a computer program product is provided, including a computer program that, when executed by a processor, implements any of the image optimization methods based on no-reference images provided in the embodiments of this application.

[0021] This application optimizes and updates the hyperparameters of the image optimization algorithm based on a no-reference image quality evaluation index, and then optimizes the image to be optimized based on the image optimization algorithm with optimized hyperparameters. This avoids the premise assumptions such as having to follow Gaussian noise distribution in the image optimization process, improves the stability of image optimization in complex image scenarios, and enhances the applicability and engineering feasibility of image optimization in complex image scenarios. Attached Figure Description

[0022] Figure 1 This is a flowchart of an image optimization method based on a no-reference image provided according to Embodiment 1 of this application;

[0023] Figure 2 This is a flowchart of an image optimization method based on a no-reference image provided according to Embodiment 2 of this application;

[0024] Figure 3 This is a schematic diagram of an image optimization device based on a no-reference image according to Embodiment 3 of this application;

[0025] Figure 4 This is a schematic diagram of the structure of an electronic device that implements the image optimization method based on a no-reference image according to Embodiment 4 of this application. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0027] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0028] Example 1

[0029] Figure 1This is a flowchart of an image optimization method based on a no-reference image, according to Embodiment 1 of this application. This embodiment is applicable to image optimization of images facing complex image scenarios such as unknown degradation kernels, unknown noise distributions, complex nonlinear priors, and multi-objective tradeoffs. It can be executed by a no-reference image optimization device, which can be implemented in hardware and / or software and can be configured in a computer device, such as a server. Figure 1 As shown, the method includes:

[0030] S110. Generate an image set to be optimized based on at least one image data to be optimized.

[0031] Here, the image to be optimized can refer to a degraded image. For example, an image may be subject to various types of degradation, such as compression, blurring, noise, and color distortion. In this embodiment of the invention, an image exhibiting degradation can be referred to as an image to be optimized.

[0032] S120. Based on a preset image optimization algorithm, perform image optimization on the image to be optimized and generate candidate images.

[0033] The image optimization algorithm may include candidate hyperparameter combinations to be optimized. It should be noted that different image optimization algorithms have different hyperparameters, such as regularization weights, penalty factors, step size / relaxation, and number of iterations. That is, the hyperparameters in the candidate hyperparameter combinations for different image optimization algorithms can be the same or different. Optionally, the image optimization algorithm can be adaptively configured according to those skilled in the art.

[0034] For example, in an embodiment of the present invention, the image optimization algorithm can be represented by the following formula:

[0035] Formula 1:

[0036] Formula 2: ;

[0037] Where x represents the candidate image; A represents the degradation model; and y represents the image to be optimized. This represents the weight of the regularization term.

[0038] The above formulas 1 and 2 are solved jointly using the alternating direction multiplier method, with custom parameters set. The process of solving for the candidate image x can be represented by the following formula:

[0039] ;

[0040] ;

[0041] ;

[0042] in, As a penalty factor; The step size is k; the number of iterations is k. It is a Lagrange multiplier.

[0043] It should be noted that in the process of solving for the target image, the candidate hyperparameter combinations can include... .

[0044] S130. Based on at least one no-reference image quality evaluation index of the candidate image, generate a target image quality score for the candidate image, and optimize and update the candidate hyperparameter combination based on the target image quality score.

[0045] Among them, the target image quality score can be used to characterize the image quality of the candidate image under the hyperparameter conditions corresponding to the candidate hyperparameter combination.

[0046] No-reference image quality assessment refers to a method of evaluating the image quality of a degraded image by analyzing its features in the absence of an original reference image. Optionally, no-reference image quality assessment metrics can be adaptively set according to those skilled in the art, such as edge information features, noise level features, and contrast and brightness features.

[0047] S140. Based on the optimized and updated target hyperparameter combination, perform image optimization on the image to be optimized again to generate the target image.

[0048] Specifically, based on the hyperparameter combination after optimization, the hyperparameters of the image optimization can be updated, and the image optimization algorithm after the hyperparameter update can be used to optimize the image again to generate the target image.

[0049] This application embodiment optimizes and updates the hyperparameters of the image optimization algorithm based on a no-reference image quality evaluation index, and then optimizes the image to be optimized based on the image optimization algorithm with optimized hyperparameters. This avoids the need to follow the premise assumptions such as Gaussian noise distribution in the image optimization process, improves the stability of image optimization in complex image scenarios, and enhances the applicability and engineering feasibility of image optimization in complex image scenarios.

[0050] Example 2

[0051] Figure 2This is a flowchart of an image optimization method based on a no-reference image, according to Embodiment 2 of this application. Based on the technical solutions of the above embodiments, this embodiment further refines the process of "generating a target image quality score for a candidate image based on at least one no-reference image quality evaluation index, and optimizing and updating the candidate hyperparameter combination based on the target image quality score." It should be noted that for parts not detailed in this embodiment, please refer to the relevant descriptions in other embodiments. Figure 2 As shown, the method includes:

[0052] S210. Generate an image set to be optimized based on at least one image data to be optimized.

[0053] S220. Based on a preset image optimization algorithm, perform image optimization on the image to be optimized and generate candidate images.

[0054] S230. For each candidate image, determine the index score of each candidate image under at least one no-reference image quality evaluation index.

[0055] Specifically, after the candidate images are generated, they can be scored according to the determined no-reference image quality evaluation index to obtain the index score of the candidate images under at least one no-reference image quality evaluation index.

[0056] S240. Determine the target image quality score of the candidate image based on the index score of the candidate image under at least one no-reference image quality evaluation index.

[0057] The target image quality score can be used to characterize the comprehensive score of a candidate image under at least one no-reference image quality evaluation index.

[0058] Optionally, based on the index scores of the candidate images under at least one no-reference image quality evaluation index, the target image quality score of the candidate images is determined, including: determining the maximum and minimum index scores of each no-reference image quality evaluation index in at least one candidate image according to the index scores of the candidate images under at least one no-reference image quality evaluation index; determining the standard score of the candidate images under each no-reference image quality evaluation index according to the maximum and minimum index scores; and performing a weighted summation of the standard scores corresponding to the no-reference image quality evaluation index according to the index weights corresponding to the no-reference image quality evaluation index to determine the target image quality score of the candidate images.

[0059] Optionally, after determining the standard score of the candidate image under each no-reference image quality evaluation index, outlier removal can be performed on the standard scores to remove abnormal extreme scores and improve data quality.

[0060] Optionally, before determining the standard scores corresponding to the no-reference image quality evaluation indicators, the process further includes: homogenizing the indicator scores corresponding to the no-reference image quality evaluation indicators. It should be noted that different no-reference image quality evaluation indicators represent image features of different dimensions, and the magnitude of the indicator values ​​of different no-reference image quality evaluation indicators represents different meanings. Therefore, it is necessary to homogenize all no-reference image quality evaluation indicators to "the larger the value, the better the quality" or "the smaller the value, the better the quality" by inverting or performing a linear transformation.

[0061] By using normalization and homogenization processes, the accuracy of the generated target image quality score is ensured.

[0062] For example, standard scores can be determined using the following formula:

[0063] ;

[0064] in, Let be the standard score corresponding to the i-th no-reference image quality evaluation index. Let i be the index score corresponding to the i-th no-reference image quality evaluation index. Let be the minimum score of the i-th no-reference image quality evaluation index on the image set to be optimized. Let be the maximum score of the i-th no-reference image quality evaluation index on the image set to be optimized.

[0065] The target image quality score can be determined using the following formula:

[0066] ;

[0067] Where Q is the target image quality score, and M is the number of no-reference image quality evaluation metrics. Let i be the weight of the no-reference image quality evaluation index. Let be the standard score corresponding to the i-th no-reference image quality evaluation index.

[0068] Optionally, before determining the score of the candidate image under at least one no-reference image quality evaluation metric, the method further includes: downsampling the candidate image.

[0069] Specifically, by downsampling the candidate images, the score of the candidate images under at least one no-reference-image quality evaluation index is determined based on the downsampling results. This reduces the feature complexity of the candidate images, improves processing efficiency, and reduces the optimization time for image optimization.

[0070] In an optional real-time mode of this invention embodiment, for the candidate image after downsampling, the standard score of the determined no-reference image quality evaluation index can be calibrated for consistency between the standard score of the candidate image before downsampling and the standard score of the candidate image after downsampling based on the monotonic mapping function of ordinal-preserving regression, thereby reducing the error between the two.

[0071] S250. If the target image quality score does not meet the preset score threshold, then based on the numerical range of the hyperparameters, the candidate hyperparameter combination is iteratively optimized until the preset iteration stop condition is met, and the target hyperparameter combination is output.

[0072] Specifically, if the target image quality score meets a preset scoring threshold, the current candidate hyperparameter combination is determined as the target hyperparameter combination. It should be noted that the preset scoring threshold can be used to control the generation of the target hyperparameter combination, and the iteration stopping condition can be used to control the iteration process of the candidate hyperparameter combinations. Optionally, the preset scoring threshold and iteration stopping condition can be adaptively set according to those skilled in the art.

[0073] Optionally, based on the numerical domain range of the hyperparameters, iterative optimization of the candidate hyperparameter combination is performed, including: for the continuous parameters in the candidate hyperparameter combination, reparameterizing each continuous parameter to generate the parameter search domain corresponding to each continuous parameter; generating a continuous parameter vector based on at least one continuous parameter in the candidate hyperparameter combination, and searching for hyperparameters in a randomly determined search direction based on the parameter search domain corresponding to each continuous parameter.

[0074] In one specific implementation, if hyperparameters The range of values ​​is The reparameterized parameter search domain (range of values) can be [-5, -1]. Other continuous parameters can also be reparameterized in the same way to determine their respective parameter search domains. After determining the parameter search domain for each continuous parameter, a continuous parameter vector can be generated for each continuous parameter. This continuous parameter vector can be used for continuous parameters in each parameter dimension, and the corresponding parameter search space is determined based on the parameter dimension of the continuous parameter vector. For example, if the continuous parameter vector includes one type of continuous parameter, the parameter search is performed in a one-dimensional space; if the continuous parameter vector includes two types of continuous parameters, the parameter search is performed in a two-dimensional space; if the continuous parameter vector includes three or more types of continuous parameters, the parameter search is performed in a higher-dimensional space of the corresponding dimension. Based on the parameter search domain of each continuous parameter, a search direction is randomly determined within the corresponding search space to sample the continuous parameter vector. For example, the search step size can be adaptively determined according to the knowledge of those skilled in the art in the search direction, and search sampling is performed in that search direction.

[0075] By reparameterizing the range of continuous parameters, numerical instability caused by hyperparameter values ​​approaching 0 during parameter search is avoided, thus improving the stability of parameter search.

[0076] Optionally, based on the numerical domain range of the hyperparameters, the iterative optimization of the candidate hyperparameter combination further includes: determining the function value change trend of the hyperparameter iterative loss function corresponding to two adjacent samples of the continuous parameter vector in the current search direction according to the hyperparameter iterative loss function; if the function value change trend is an upward trend, then the search direction of the continuous parameter vector is updated.

[0077] By updating the search direction, the search accuracy of the hyperparameter search is ensured.

[0078] If a preset iteration stop condition is met during the parameter search process, the parameter search will stop. For example, if the change trend of the hyperparameter iteration loss function corresponding to two adjacent samples of the continuous parameter vector in the current search direction is that the loss function value decreases, and the difference between the loss function values ​​of the hyperparameter iteration loss function corresponding to two consecutively sampled continuous parameter vectors is less than the preset loss function value, or the number of parameter searches meets the preset search number threshold, then the parameter search will stop and the current continuous parameters will be output.

[0079] It should be noted that the hyperparameter iterative loss function can be used to indicate the search performance of the search direction, and can be adaptively set according to those skilled in the art.

[0080] Optionally, in one implementation, for discrete parameters in the candidate hyperparameter combination, such as the number of iterations, if the number of iterations k belongs to {20, 40, 80, 120}, k can be fixed to a medium value (such as 40) within the range of values ​​by outer enumeration. Under this condition, the continuous parameters in the candidate hyperparameter combination can be optimized, and finally the continuous parameters corresponding to different number of iterations k can be determined.

[0081] S260. Based on the optimized and updated target hyperparameter combination, perform image optimization on the image to be optimized again to generate the target image.

[0082] This application embodiment achieves image processing tasks with simultaneous optimization of multiple parameters by iteratively optimizing candidate hyperparameter combinations based on index scores under a no-reference image quality evaluation index. This overcomes the technical problems of traditional parameter search relying on human experience or grid search, which is time-consuming and difficult to extend to multi-parameter optimization scenarios, thus improving the optimization performance of image optimization.

[0083] Example 3

[0084] Figure 3 This is a schematic diagram of an image optimization device based on a no-reference image, according to Embodiment 3 of this application. It is applicable to image optimization in complex image scenarios such as unknown degradation kernels, unknown noise distributions, complex nonlinear priors, and multi-objective tradeoffs. This no-reference image optimization device can be implemented in hardware and / or software and can be configured in a computer device, such as a server. Figure 3 As shown, the device includes:

[0085] Image set generation module 310 is used to generate an image set to be optimized based on at least one image data to be optimized; wherein, the image to be optimized refers to a degraded image;

[0086] The candidate image generation module 320 is used to optimize the image to be optimized based on a preset image optimization algorithm to generate candidate images; wherein, the image optimization algorithm includes candidate hyperparameter combinations to be optimized;

[0087] The hyperparameter optimization module 330 is used to generate a target image quality score for the candidate image based on at least one no-reference image quality evaluation index of the candidate image, and to optimize and update the candidate hyperparameter combination based on the target image quality score; wherein, the target image quality score is used to characterize the image quality of the candidate image under the hyperparameter conditions corresponding to the candidate hyperparameter combination.

[0088] The target image generation module 340 is used to perform image optimization on the image to be optimized again and generate the target image based on the optimized and updated target hyperparameter combination.

[0089] This application embodiment optimizes and updates the hyperparameters of the image optimization algorithm based on a no-reference image quality evaluation index, and then optimizes the image to be optimized based on the image optimization algorithm with optimized hyperparameters. This avoids the need to follow the premise assumptions such as Gaussian noise distribution in the image optimization process, improves the stability of image optimization in complex image scenarios, and enhances the applicability and engineering feasibility of image optimization in complex image scenarios.

[0090] Optionally, the hyperparameter optimization module 330 includes:

[0091] The index score determination unit is used to determine the index score of each candidate image under at least one no-reference image quality evaluation index for the candidate images.

[0092] The scoring determination unit is used to determine the target image quality score of the candidate image based on the index score of the candidate image under at least one no-reference image quality evaluation index; wherein the target image quality score is used to characterize the comprehensive score of the candidate image under at least one no-reference image quality evaluation index.

[0093] The parameter optimization unit is used to iteratively optimize the candidate hyperparameter combination based on the numerical range of the hyperparameters if the target image quality score does not meet the preset score threshold, until the preset iteration stop condition is met, and output the target hyperparameter combination.

[0094] Optionally, the parameter optimization unit includes:

[0095] The reparameter subunit is used to reparameterize each continuous parameter in the candidate hyperparameter combination, generating the parameter search domain corresponding to each continuous parameter.

[0096] The parameter optimization subunit is used to generate a continuous parameter vector based on at least one continuous parameter in the candidate hyperparameter combination, and to search for hyperparameters in a randomly determined search direction based on the parameter search domain corresponding to each continuous parameter.

[0097] Optionally, the parameter optimization unit may also include:

[0098] The direction update subunit is used to determine the function value change trend of the hyperparameter iterative loss function corresponding to two adjacent samples of the continuous parameter vector in the current search direction based on the hyperparameter iterative loss function; if the function value change trend is an upward trend, the search direction of the continuous parameter vector is updated.

[0099] Optionally, the scoring unit can be specifically used for:

[0100] Based on the index scores of the candidate images under at least one no-reference image quality evaluation index, determine the maximum index score and minimum index score of each no-reference image quality evaluation index in at least one candidate image.

[0101] Based on the maximum and minimum index scores, determine the standard score of the candidate image for each no-reference image quality evaluation index;

[0102] Based on the index weights corresponding to the no-reference image quality evaluation index, the standard scores corresponding to the no-reference image quality evaluation index are weighted and summed to determine the target image quality score of the candidate image.

[0103] Optionally, the scoring determination unit may also include:

[0104] A downsampling unit is used to downsample the candidate image before determining the index score of the candidate image under at least one no-reference image quality evaluation index.

[0105] The image optimization apparatus based on no-reference images provided in this application can execute the image optimization method based on no-reference images provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects for executing each image optimization method based on no-reference images.

[0106] According to embodiments of this application, this application also provides an electronic device, a readable storage medium, and a computer program product.

[0107] Example 4

[0108] Figure 4 This is a schematic diagram of the structure of an electronic device 410 implementing the image optimization method based on a referenceless image according to embodiments of this application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or claimed herein.

[0109] like Figure 4 As shown, the electronic device 410 includes at least one processor 411 and a memory, such as a read-only memory 412 or a random access memory 413, communicatively connected to the at least one processor 411. The memory stores computer programs executable by the at least one processor. The processor 411 can perform various appropriate actions and processes based on the computer program stored in the read-only memory 412 or loaded from storage unit 418 into the random access memory 413. The random access memory 413 can also store various programs and data required for the operation of the electronic device 410. The processor 411, read-only memory 412, and random access memory 413 are interconnected via a bus 414. An input / output interface 415 is also connected to the bus 414.

[0110] Multiple components in electronic device 410 are connected to input / output interface 415, including: input unit 416, such as keyboard, mouse, etc.; output unit 417, such as various types of monitors, speakers, etc.; storage unit 418, such as disk, optical disk, etc.; and communication unit 419, such as network card, modem, wireless transceiver, etc. Communication unit 419 allows electronic device 410 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0111] Processor 411 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 411 include, but are not limited to, central processing units, graphics processing units, various special-purpose artificial intelligence computing chips, various processors running machine learning model algorithms, digital signal processors, and any suitable processor, controller, microcontroller, etc. Processor 411 performs the various methods and processes described above, such as image optimization methods based on no-reference images.

[0112] In some embodiments, the image optimization method based on a no-reference image can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 418. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 410 via read-only memory 412 and / or communication unit 419. When the computer program is loaded into random access memory 413 and executed by processor 411, one or more steps of the image optimization method based on a no-reference image described above can be performed. Alternatively, in other embodiments, processor 411 can be configured for the image optimization method based on a no-reference image by any other suitable means (e.g., by means of firmware).

[0113] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays, application-specific integrated circuits (ASICs), application-specific standard products (ASICs), systems-on-a-chip (SoCs), payload programmable logic devices, computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0114] Computer programs used to implement the methods of this application may be written in any combination of one or more programming languages. These computer programs may be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable image optimization device based on a referenceless image, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0115] In the context of this application, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0116] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a cathode ray tube or liquid crystal display monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0117] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0118] A computing system can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product within the cloud computing service system to address the shortcomings of traditional physical hosts and virtual private servers, such as high management difficulty and weak business scalability.

[0119] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this application can be achieved, and this is not limited herein.

[0120] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. An image optimization method based on a no-reference image, characterized in that, include: A set of images to be optimized is generated based on at least one image data to be optimized; wherein, the image to be optimized refers to a degraded image; Based on a preset image optimization algorithm, image optimization is performed on the image to be optimized to generate candidate images; wherein, the image optimization algorithm includes candidate hyperparameter combinations to be optimized; Based on at least one no-reference image quality evaluation index of the candidate image, a target image quality score is generated for the candidate image, and the candidate hyperparameter combination is optimized and updated based on the target image quality score; wherein, the target image quality score is used to characterize the image quality of the candidate image under the hyperparameter conditions corresponding to the candidate hyperparameter combination. Based on the optimized and updated combination of target hyperparameters, the image to be optimized is optimized again to generate the target image.

2. The method according to claim 1, characterized in that, The process of generating a target image quality score for the candidate image based on at least one no-reference image quality evaluation metric, and optimizing and updating the candidate hyperparameter combination based on the target image quality score, includes: For each candidate image, determine the index score of each candidate image under at least one no-reference image quality evaluation index; Based on the index scores of the candidate image under at least one no-reference image quality evaluation index, the target image quality score of the candidate image is determined; wherein, the target image quality score is used to characterize the comprehensive score of the candidate image under at least one no-reference image quality evaluation index. If the target image quality score does not meet the preset score threshold, the candidate hyperparameter combination is iteratively optimized based on the numerical range of the hyperparameters until the preset iteration stop condition is met, and the target hyperparameter combination is output.

3. The method according to claim 2, characterized in that, The iterative optimization of candidate hyperparameter combinations based on the numerical domain range of the hyperparameters includes: For continuous parameters in the candidate hyperparameter combination, each continuous parameter is reparameterized to generate the parameter search domain corresponding to each continuous parameter. Based on at least one continuous parameter in the candidate hyperparameter combination, a continuous parameter vector is generated, and a hyperparameter search is performed in a randomly determined search direction based on the parameter search domain corresponding to each continuous parameter.

4. The method according to claim 3, characterized in that, Also includes: Based on the hyperparameter iterative loss function, determine the trend of the function value change of the hyperparameter iterative loss function corresponding to the continuous parameter vectors sampled in two adjacent samples in the current search direction; If the function value changes in an upward trend, the search direction of the continuous parameter vector is updated.

5. The method according to claim 2, characterized in that, The step of determining the target image quality score of the candidate image based on the index score of the candidate image under at least one no-reference image quality evaluation index includes: Based on the index scores of the candidate images under at least one no-reference image quality evaluation index, determine the maximum index score and minimum index score of each no-reference image quality evaluation index in at least one candidate image. Based on the maximum and minimum index scores, determine the standard score of the candidate image for each no-reference image quality evaluation index; Based on the index weights corresponding to the no-reference image quality evaluation index, the standard scores corresponding to the no-reference image quality evaluation index are weighted and summed to determine the target image quality score of the candidate image.

6. The method according to claim 5, characterized in that, Before determining the score of the candidate image under at least one no-reference image quality metric, the following steps are also included: The candidate image is downsampled.

7. An image optimization apparatus based on a no-reference image, characterized in that, include: An image set generation module is used to generate an image set to be optimized based on at least one image data to be optimized; wherein, the image to be optimized refers to a degraded image; A candidate image generation module is used to optimize an image to be optimized based on a preset image optimization algorithm to generate candidate images; wherein, the image optimization algorithm includes candidate hyperparameter combinations to be optimized; The hyperparameter optimization module is used to generate a target image quality score for the candidate image based on at least one no-reference image quality evaluation index, and to optimize and update the candidate hyperparameter combination based on the target image quality score; wherein, the target image quality score is used to characterize the image quality of the candidate image under the hyperparameter conditions corresponding to the candidate hyperparameter combination. The target image generation module is used to perform image optimization on the image to be optimized again and generate the target image based on the optimized and updated target hyperparameter combination.

8. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the image optimization method based on a no-reference image as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the image optimization method based on a no-reference image as described in any one of claims 1-6.

10. A computer program product comprising a computer program that, when executed by a processor, implements the image optimization method based on a no-reference image according to any one of claims 1-6.

Images