Intelligent adjustment method and device for image quality parameters

Abstract

The application discloses an intelligent image quality parameter adjusting method and device, and the method comprises the following steps: identifying an image through a recognition model to obtain an image recognition result; determining an image information set according to the image recognition result; determining an image quality parameter corresponding to the image according to the scene information set; and adjusting the current image quality parameter of the image according to the image quality parameter. It can be seen that, after the image is identified through the recognition model, the global image quality parameter is adjusted according to the directly obtained scene category confidence, and the local image quality parameter is adjusted according to the directly obtained skin color confidence of the face and non-face area, so that the problem that the scene category recognition is limited in the traditional image quality parameter adjusting method is solved, the overall image quality parameter accuracy is improved through the image local area (face and skin color area) quality parameter adjustment, the overall image quality effect is enhanced, and the user's visual experience is improved.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to an intelligent adjustment method and apparatus for image quality parameters. Background Technology

[0002] With the rapid development of smart display devices, the image display quality of these devices is becoming increasingly higher. Excellent image display quality allows viewers to fully appreciate the image's clarity, color richness, three-dimensionality, and color comfort, thus enabling them to enjoy the stunning visual effects brought by smart display devices.

[0003] Currently, improving image display quality generally requires adjusting image quality parameters (such as contrast and brightness). A common method for adjusting these parameters involves matching identified scene categories in the image with a pre-stored library of real-world scene categories, calculating the scene confidence score for each category, and then adjusting the image quality parameters based on this confidence score to improve display quality. However, practical experience has shown that traditional methods have limitations in identifying scene categories. For some specific scene categories, matching with the real-world scene category library is difficult, making it hard to precisely adjust image quality parameters based on the scene confidence score, thus failing to effectively improve display quality. Therefore, providing a method for precisely adjusting image quality parameters is crucial. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and apparatus for intelligent adjustment of image quality parameters, which can solve the problem of limited image scene category recognition in traditional image quality adjustment methods, and can also improve the accuracy of image quality adjustment of the whole image by adjusting the image quality parameters of local areas of the image (face and skin color areas), thereby enhancing the overall image quality effect and improving the user's visual experience.

[0005] To address the aforementioned technical problems, the first aspect of this invention discloses an intelligent adjustment method for image quality parameters, the method comprising:

[0006] The target recognition model performs image recognition operations on the target image to be recognized, and obtains the image recognition result corresponding to the target image; the target recognition model includes a scene recognition model, a face detection model, and a skin color detection model;

[0007] Based on the image recognition results, an image information set for the target image is determined; the image information set for the target image includes a scene information set for the target image, and each scene information in the scene information set includes a scene confidence score corresponding to each scene category;

[0008] Based on the scene information set, determine the first image quality parameters that match the target image;

[0009] Based on the first image quality parameters, a first adjustment operation is performed on the current image quality parameters of the target image.

[0010] As an optional implementation, in the first aspect of the present invention, each scene category has a corresponding preset scene confidence threshold group, and the preset scene confidence threshold group corresponding to each scene category includes the maximum scene confidence threshold and the minimum scene confidence threshold corresponding to that scene category.

[0011] The step of determining the first image quality parameter matching the target image based on the scene information set includes:

[0012] Based on the scene confidence level corresponding to each scene category, a first undetermined scene category is determined from all the scene categories; the first undetermined scene category is the scene category with the highest scene confidence level among all the scene categories;

[0013] Analyze the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence threshold group corresponding to the first undetermined scene category;

[0014] Based on the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence threshold group corresponding to the first undetermined scene category, a first image quality parameter matching the target image is determined.

[0015] As an optional implementation, in the first aspect of the present invention, determining the first image quality parameter matching the target image based on a comparison between the scene confidence level corresponding to the first undetermined scene category and a preset scene confidence threshold group corresponding to the first undetermined scene category includes:

[0016] When the scene confidence level corresponding to the first undetermined scene category is less than the minimum scene confidence threshold corresponding to the first undetermined scene category, the preset image quality parameter is determined as the first image quality parameter that matches the target image.

[0017] When the scene confidence level corresponding to the first undetermined scene category is greater than or equal to the maximum scene confidence level threshold corresponding to the first undetermined scene category, the first image quality parameter matching the target image is determined based on the scene confidence level corresponding to the first undetermined scene category.

[0018] When the scene confidence level corresponding to the first undetermined scene category is greater than or equal to the minimum scene confidence level threshold corresponding to the first undetermined scene category and less than the maximum scene confidence level threshold corresponding to the first undetermined scene category, all second undetermined scene categories other than the first undetermined scene category are determined from all the scene categories; the scene confidence level corresponding to each second undetermined scene category is greater than or equal to the minimum scene confidence level corresponding to that second undetermined scene category;

[0019] Calculate the sum of the scene confidence scores corresponding to the first undetermined scene category and the scene confidence scores corresponding to each of the second undetermined scene categories;

[0020] From all the second undetermined scenario categories, determine all the second undetermined scenario categories whose sum of scenario confidence is greater than or equal to a preset mixed scenario confidence threshold, and use them as all third undetermined scenario categories;

[0021] Based on the scene confidence scores corresponding to the first undetermined scene category and all the scene confidence scores corresponding to the third undetermined scene categories, a first image quality parameter matching the target image is determined.

[0022] As an optional implementation, in the first aspect of the present invention, determining the first image quality parameter matching the target image based on the scene confidence level corresponding to the first undetermined scene category and the scene confidence levels corresponding to all the third undetermined scene categories includes:

[0023] For the first undetermined scene category and all the third undetermined scene categories, a scene feature extraction operation is performed on the target image to obtain the first scene feature corresponding to the first undetermined scene category and the second scene feature corresponding to each of the third undetermined scene categories;

[0024] Based on the first scene features and all the second scene features, an adjustment operation is performed on the scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories to obtain the adjusted scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories.

[0025] Based on the adjusted scene confidence scores corresponding to the first undetermined scene category and the scene confidence scores corresponding to all the third undetermined scene categories, a first image quality parameter matching the target image is determined.

[0026] As an optional implementation, in a first aspect of the invention, before performing a first adjustment operation on the current image quality parameters of the target image according to the first image quality parameters, the method further includes:

[0027] Determine whether the image information set of the target image also includes the target region information set of the target image; the target region information set includes a skin color region information set, or the skin color region information set and a face region information set, wherein each skin color region information in the skin color region information set includes a first skin color confidence level for each skin color region, and each face region information in the face region information set includes a second skin color confidence level for each face region;

[0028] When it is determined that the image information set of the target image does not include the target region information set of the target image, the step of performing the first adjustment operation on the current image quality parameter of the target image based on the first image quality parameter is triggered.

[0029] When it is determined that the image information set of the target image also includes the target region information set of the target image, a second image quality parameter matching the target image is determined based on the target region information set.

[0030] Based on the first image quality parameters and the second image quality parameters, a second adjustment operation is performed on the current image quality parameters of the target image.

[0031] As an optional implementation, in the first aspect of the present invention, determining the second image quality parameter matching the target image based on the target region information set includes:

[0032] When the target region information set only includes the skin color region information set, based on the first skin color confidence of all the skin color regions, all skin color regions whose first skin color confidence is greater than or equal to a preset first confidence threshold are determined from all the skin color regions and are taken as all first target skin color regions;

[0033] Based on the first skin color confidence scores of all the first target skin color regions, determine the second image quality parameters that match the target image;

[0034] When the target region information set includes the skin color region information set and the face region information set, a second image quality parameter matching the target image is determined based on the first skin color confidence score of all the skin color regions and the second skin color confidence score of all the face regions.

[0035] As an optional implementation, in a first aspect of the invention, determining the second image quality parameter matching the target image based on the first skin color confidence level of all said skin color regions and the second skin color confidence level of all said face regions includes:

[0036] Based on all the face regions, determine all first regions to be processed that overlap with the corresponding face regions from all the skin color regions, and determine all second regions to be processed that do not overlap with the corresponding face regions from all the skin color regions.

[0037] Based on the location information of each first region to be processed, a first weight and a second weight of each second region to be processed are determined; the location information of each first region to be processed includes the distance between the first region to be processed and all the second regions to be processed.

[0038] According to the first weight of each first region to be processed, an adjustment operation is performed on the second skin color confidence of the face region corresponding to each first region to be processed to update the second skin color confidence of all face regions; and according to the second weight of each second region to be processed, an adjustment operation is performed on the first skin color confidence of the skin color region corresponding to each second region to be processed to update the first skin color confidence of all skin color regions.

[0039] After updating the first skin color confidence of all skin color regions and the second skin color confidence of all face regions, based on the first skin color confidence of all skin color regions, all skin color regions whose first skin color confidence is greater than or equal to the preset first confidence threshold are determined from all skin color regions as all second target skin color regions; and based on the second skin color confidence of all face regions, all face regions whose second skin color confidence is greater than or equal to the preset second confidence threshold are determined from all face regions as all target face regions.

[0040] Based on the first skin color confidence scores of all the second target skin color regions and the second skin color confidence scores of all the target face regions, a second image quality parameter matching the target image is determined.

[0041] A second aspect of the present invention discloses an intelligent adjustment device for image quality parameters, the device comprising:

[0042] The image recognition module is used to perform image recognition operations on the target image to be recognized through a target recognition model, and obtain the image recognition result corresponding to the target image; the target recognition model includes a scene recognition model, a face detection model, and a skin color detection model.

[0043] The first determining module is used to determine the image information set of the target image based on the image recognition result; the image information set of the target image includes the scene information set of the target image, and each scene information in the scene information set includes the scene confidence level corresponding to each scene category;

[0044] The second determining module is used to determine a first image quality parameter that matches the target image based on the scene information set;

[0045] The adjustment module is used to perform a first adjustment operation on the current image quality parameters of the target image based on the first image quality parameters.

[0046] As an optional implementation, in the second aspect of the present invention, each scene category has a corresponding preset scene confidence threshold group, and the preset scene confidence threshold group corresponding to each scene category includes the maximum scene confidence threshold and the minimum scene confidence threshold corresponding to that scene category.

[0047] The second determining module includes:

[0048] The first determining submodule is used to determine a first undetermined scene category from all the scene categories based on the scene confidence level corresponding to each scene category; the first undetermined scene category is the scene category with the highest scene confidence level among all the scene categories;

[0049] The analysis submodule is used to analyze the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence level threshold group corresponding to the first undetermined scene category.

[0050] The second determining submodule is used to determine the first image quality parameter that matches the target image based on the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence threshold group corresponding to the first undetermined scene category.

[0051] As an optional implementation, in a second aspect of the invention, the second determining submodule includes:

[0052] The determining unit is configured to: when the scene confidence level corresponding to the first undetermined scene category is less than the minimum scene confidence level threshold corresponding to the first undetermined scene category, determine a preset image quality parameter as a first image quality parameter matching the target image; when the scene confidence level corresponding to the first undetermined scene category is greater than or equal to the maximum scene confidence level threshold corresponding to the first undetermined scene category, determine a first image quality parameter matching the target image based on the scene confidence level corresponding to the first undetermined scene category; when the scene confidence level corresponding to the first undetermined scene category is greater than or equal to the minimum scene confidence level threshold corresponding to the first undetermined scene category and less than the maximum scene confidence level threshold corresponding to the first undetermined scene category, determine all second undetermined scene categories other than the first undetermined scene category from all the scene categories; the scene confidence level corresponding to each second undetermined scene category is greater than or equal to the minimum scene confidence level threshold corresponding to that second undetermined scene category;

[0053] The calculation unit is used to calculate the sum of the scene confidence scores corresponding to the first undetermined scene category and the scene confidence scores corresponding to each of the second undetermined scene categories;

[0054] The determining unit is further configured to determine all second undetermined scene categories from all second undetermined scene categories whose sum of scene confidence is greater than or equal to a preset mixed scene confidence threshold, as all third undetermined scene categories; and determine a first image quality parameter that matches the target image based on the scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories.

[0055] As an optional implementation, in a second aspect of the present invention, the determining unit determines the first image quality parameter matching the target image based on the scene confidence level corresponding to the first undetermined scene category and the scene confidence levels corresponding to all the third undetermined scene categories in the following specific manner:

[0056] For the first undetermined scene category and all the third undetermined scene categories, a scene feature extraction operation is performed on the target image to obtain the first scene feature corresponding to the first undetermined scene category and the second scene feature corresponding to each of the third undetermined scene categories;

[0057] Based on the first scene features and all the second scene features, an adjustment operation is performed on the scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories to obtain the adjusted scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories.

[0058] Based on the adjusted scene confidence scores corresponding to the first undetermined scene category and the scene confidence scores corresponding to all the third undetermined scene categories, a first image quality parameter matching the target image is determined.

[0059] As an optional implementation, in a second aspect of the invention, the apparatus further includes:

[0060] The judgment module is configured to determine, before the adjustment module performs a first adjustment operation on the current image quality parameters of the target image based on the first image quality parameters, whether the image information set of the target image still includes the target region information set of the target image; the target region information set includes a skin color region information set, or the skin color region information set and a face region information set, wherein each skin color region information in the skin color region information set includes a first skin color confidence level for each skin color region, and each face region information in the face region information set includes a second skin color confidence level for each face region; when it is determined that the image information set of the target image does not include the target region information set of the target image, the adjustment module is triggered to perform the step of performing a first adjustment operation on the current image quality parameters of the target image based on the first image quality parameters;

[0061] The third determining module is used to determine a second image quality parameter that matches the target image based on the target region information set when the determining module determines that the image information set of the target image also includes the target region information set of the target image.

[0062] The adjustment module is further configured to perform a second adjustment operation on the current image quality parameters of the target image based on the first image quality parameters and the second image quality parameters.

[0063] As an optional implementation, in a second aspect of the invention, the third determining module includes:

[0064] The third determining submodule is used to: when the target region information set only includes the skin color region information set, determine all skin color regions whose first skin color confidence is greater than or equal to a preset first confidence threshold, based on the first skin color confidence of all skin color regions, as all first target skin color regions; determine a second image quality parameter matching the target image based on the first skin color confidence of all first target skin color regions; and when the target region information set includes the skin color region information set and the face region information set, determine a second image quality parameter matching the target image based on the first skin color confidence of all skin color regions and the second skin color confidence of all face regions.

[0065] As an optional implementation, in the second aspect of the present invention, the third determining submodule determines the second image quality parameters matching the target image based on the first skin color confidence scores of all the skin color regions and the second skin color confidence scores of all the face regions in the following specific manner:

[0066] Based on all the face regions, determine all first regions to be processed that overlap with the corresponding face regions from all the skin color regions, and determine all second regions to be processed that do not overlap with the corresponding face regions from all the skin color regions.

[0067] Based on the location information of each first region to be processed, a first weight and a second weight of each second region to be processed are determined; the location information of each first region to be processed includes the distance between the first region to be processed and all the second regions to be processed.

[0068] According to the first weight of each first region to be processed, an adjustment operation is performed on the second skin color confidence of the face region corresponding to each first region to be processed to update the second skin color confidence of all face regions; and according to the second weight of each second region to be processed, an adjustment operation is performed on the first skin color confidence of the skin color region corresponding to each second region to be processed to update the first skin color confidence of all skin color regions.

[0069] After updating the first skin color confidence of all skin color regions and the second skin color confidence of all face regions, based on the first skin color confidence of all skin color regions, all skin color regions whose first skin color confidence is greater than or equal to the preset first confidence threshold are determined from all skin color regions as all second target skin color regions; and based on the second skin color confidence of all face regions, all face regions whose second skin color confidence is greater than or equal to the preset second confidence threshold are determined from all face regions as all target face regions.

[0070] Based on the first skin color confidence scores of all the second target skin color regions and the second skin color confidence scores of all the target face regions, a second image quality parameter matching the target image is determined.

[0071] A third aspect of the present invention discloses another intelligent adjustment device for image quality parameters, the device comprising:

[0072] Memory containing executable program code;

[0073] A processor coupled to the memory;

[0074] The processor calls the executable program code stored in the memory to execute the intelligent adjustment method for image quality parameters disclosed in the first aspect of the present invention.

[0075] The fourth aspect of the present invention discloses a computer-storable medium storing computer instructions, which, when invoked, are used to execute the intelligent adjustment method for image quality parameters disclosed in the first aspect of the present invention.

[0076] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0077] In this embodiment of the invention, an image recognition operation is performed on the target image to be recognized using a target recognition model to obtain the image recognition result corresponding to the target image; based on the image recognition result, the image information set of the target image is determined; based on the scene information set, a first image quality parameter matching the target image is determined; based on the first image quality parameter, a first adjustment operation is performed on the current image quality parameter of the target image. It can be seen that implementing this invention enables global image quality adjustment based on the directly obtained scene category confidence score after image recognition by the recognition model, and also enables local image quality adjustment based on the directly obtained skin color confidence score of face and non-face regions. This eliminates the need for scene feature matching with a real-world scene category library to obtain scene category confidence scores. Thus, it not only solves the problem of limited image scene category recognition in traditional image quality adjustment methods, but also improves the accuracy of overall image quality adjustment by adjusting the image quality of local areas (face and skin color regions), thereby enhancing the overall image quality effect and improving the user's visual experience. Attached Figure Description

[0078] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0079] Figure 1 This is a flowchart illustrating an intelligent adjustment method for image quality parameters disclosed in an embodiment of the present invention.

[0080] Figure 2 This is a flowchart illustrating another intelligent adjustment method for image quality parameters disclosed in an embodiment of the present invention;

[0081] Figure 3 This is a schematic diagram of the structure of an intelligent adjustment device for image quality parameters disclosed in an embodiment of the present invention;

[0082] Figure 4 This is a schematic diagram of another intelligent adjustment device for image quality parameters disclosed in an embodiment of the present invention;

[0083] Figure 5 This is a schematic diagram of the structure of another intelligent adjustment device for image quality parameters disclosed in an embodiment of the present invention. Detailed Implementation

[0084] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0085] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0086] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0087] This invention discloses an intelligent adjustment method and apparatus for image quality parameters, which can solve the problem of limited image scene category recognition in traditional image quality adjustment methods. Furthermore, by adjusting the image quality parameters of local areas (faces and skin tones), it can improve the accuracy of overall image quality adjustment, thereby enhancing the overall image quality and improving the user's visual experience. Detailed descriptions follow.

[0088] Example 1

[0089] Please see Figure 1 , Figure 1 This is a flowchart illustrating an intelligent adjustment method for image quality parameters disclosed in an embodiment of the present invention. Figure 1The described intelligent adjustment method for image quality parameters can be applied to adjusting the image quality parameters of smart display devices such as smart TVs and smartphones, and also to adjusting the image quality parameters of network service products such as live video and video conferencing. This invention does not limit the application of this method. Optionally, this method can be implemented by an image processing system, which can be integrated into an image processing device. This system can be a local server or a cloud server for managing the image processing (parameter adjustment) workflow, and this invention does not limit the application of this method. Figure 1 As shown, the intelligent adjustment method for image quality parameters can include the following operations:

[0090] 101. Using the target recognition model, perform image recognition operations on the target image to be recognized to obtain the image recognition result corresponding to the target image.

[0091] In this embodiment of the invention, the target recognition model includes a scene recognition model, a face detection model, and a skin color detection model. Further, performing an image recognition operation on the target image to be recognized using the target recognition model to obtain the image recognition result corresponding to the target image may include: performing a first recognition operation on the target image to be recognized using a first recognition model to obtain a first recognition result of the target image, and performing a second recognition operation on the target image using a second recognition model to obtain a second recognition result of the target image; the first recognition model includes a scene recognition model and a face detection model, and the second recognition model includes a skin color detection model; the first recognition result and the second recognition result of the target image are determined as the image recognition result corresponding to the target image.

[0092] 102. Based on the image recognition results, determine the set of image information for the target image.

[0093] In this embodiment of the invention, the image information set of the target image includes a scene information set of the target image, and each scene information in the scene information set includes a scene confidence score corresponding to each scene category. Optionally, the scene category can be blue sky, grass, building, person, fruit, flower, food, mountain, snow scene, etc. Specifically, the image recognition result can be understood as the recognition result related to scene, face, and skin color obtained after the scene recognition model, face detection model, and skin color detection model recognize the target image, such as the confidence score corresponding to the scene category, scene region location, number of faces, face region location, skin color confidence score corresponding to the face region, skin color region location, skin color confidence score corresponding to the skin color region, non-skin color region location, etc. The recognition results related to face and skin color may not be recognized, because there is no face region or skin color region in the target image, or the size of the face region and skin color region is too small. Further, if the recognition results related to face and skin color can be recognized, the face region information and skin color region information of the target image can be determined based on the image recognition results.

[0094] 103. Based on the scene information set, determine the first image quality parameters that match the target image.

[0095] In this embodiment of the invention, specifically, based on the scene information set, global region image quality parameters matching the target image can be determined. If the recognition results related to the face and skin color can be recognized, then local region (face region and skin color region) image quality parameters matching the target image can also be determined based on the face region information and skin color region information.

[0096] 104. Based on the first image quality parameters, perform the first adjustment operation on the current image quality parameters of the target image.

[0097] In this embodiment of the invention, the first image quality parameter can be understood as preset brightness, preset contrast, preset saturation, preset hue, preset color temperature, and preset clarity, etc., corresponding to the scene category in the target image, while the current image quality parameter of the target image can be understood as the current brightness, current contrast, current saturation, current hue, current color temperature, and current clarity, etc. of the target image.

[0098] As can be seen, implementing the embodiments of the present invention enables global image quality tuning based on the directly obtained scene category confidence score after image recognition by the recognition model. It also enables local image quality tuning based on the directly obtained skin color confidence score of face and non-face regions. This eliminates the need for scene feature matching with a real-world scene category library to obtain scene category confidence score. In this way, it not only solves the problem of limited image scene category recognition in traditional image quality tuning methods, but also improves the accuracy of overall image quality tuning by tuning local areas of the image (face and skin color regions), thereby enhancing the overall image quality effect and improving the user's visual experience.

[0099] In an optional embodiment, step 103 above, determining the first image quality parameter matching the target image based on the scene information set, includes:

[0100] Based on the scene confidence level corresponding to each scene category, the first undetermined scene category is determined from all scene categories;

[0101] Analyze the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence threshold group corresponding to the first undetermined scene category;

[0102] Based on the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence threshold group corresponding to the first undetermined scene category, the first image quality parameter that matches the target image is determined.

[0103] In this optional embodiment, each scene category has a corresponding preset scene confidence threshold group, and the preset scene confidence threshold group for each scene category includes the maximum scene confidence threshold and the minimum scene confidence threshold for that scene category. The first scene category to be determined is the scene category with the highest scene confidence among all scene categories. Specifically, the scene confidence corresponding to each scene category can be understood as the probability that the scene category exists in the target image. The preset scene confidence threshold groups for different scene categories can be different from each other. By comparing the scene confidence of the scene category with its own preset scene confidence threshold group, it can be determined whether the target image contains the scene category, a mixed scene of the scene category, or does not belong to any of the existing scene categories. For example, in the target image, the scene category with the highest scene confidence is blue sky, and its corresponding preset scene confidence threshold group is [0.3, 0.65]. When the scene confidence corresponding to blue sky is 0.9, it can be considered that only the blue sky scene exists in the target image; when the scene confidence corresponding to blue sky is 0.5, it can be considered that not only the blue sky scene exists in the target image; and when the scene confidence corresponding to blue sky is 0.2, it can be considered that none of the existing scene categories exist in the target image.

[0104] As can be seen, this optional embodiment can determine the scene category in the target image by comparing the confidence level corresponding to the scene category with its own preset scene confidence threshold group. At the same time, it can adjust the image quality parameters of the target image by using the first undetermined scene category with the maximum scene confidence. This helps to reduce the impact of limited scene category recognition, improve the range of scene categories that the image processing system can recognize, and thus improve the reliability and accuracy of the overall image quality parameter adjustment of the target image, thereby improving the image quality effect of the target image and the viewer's visual experience.

[0105] In another optional embodiment, the step of determining the first image quality parameter matching the target image based on the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence threshold group corresponding to the first undetermined scene category includes:

[0106] When the scene confidence level corresponding to the first undetermined scene category is less than the minimum scene confidence threshold corresponding to the first undetermined scene category, the preset image quality parameter is determined as the first image quality parameter that matches the target image.

[0107] When the scene confidence level corresponding to the first undetermined scene category is greater than or equal to the maximum scene confidence level threshold corresponding to the first undetermined scene category, the first image quality parameter that matches the target image is determined based on the scene confidence level corresponding to the first undetermined scene category.

[0108] When the scene confidence level corresponding to the first undetermined scene category is greater than or equal to the minimum scene confidence level threshold corresponding to the first undetermined scene category and less than the maximum scene confidence level threshold corresponding to the first undetermined scene category, all second undetermined scene categories other than the first undetermined scene category are determined from all scene categories.

[0109] Calculate the sum of the scene confidence scores corresponding to the first undetermined scene category and the scene confidence scores corresponding to each of the second undetermined scene categories;

[0110] From all the second undetermined scenario categories, identify all the second undetermined scenario categories whose sum of scenario confidence is greater than or equal to a preset mixed scenario confidence threshold, and use them as all third undetermined scenario categories;

[0111] Based on the scene confidence scores corresponding to the first undetermined scene category and all the scene confidence scores corresponding to the third undetermined scene categories, determine the first image quality parameters that match the target image.

[0112] In this optional embodiment, the scene confidence score corresponding to each second undetermined scene category is greater than or equal to the minimum scene confidence score threshold corresponding to that second undetermined scene category. For example, for scene category A with the highest scene confidence score, when its corresponding scene confidence score is less than its own minimum scene confidence score threshold, that is, there are no existing scene categories in the target image, the default image quality parameters can be set to image quality parameters matching the target image; while when its corresponding scene confidence score is greater than or equal to its own maximum scene confidence score threshold, that is, it can be considered that only scene category A exists in the target image, and the image quality parameters matching the target image can be directly set based on the scene confidence score corresponding to scene category A; When the scene confidence score of a scene is greater than or equal to its minimum scene confidence threshold and less than its maximum scene confidence threshold, it can be assumed that the target image contains not only scene category A but also other scene categories. In this case, it is necessary to first filter out all scene categories B (i.e., scene categories whose scene confidence score is greater than their minimum scene confidence threshold) from all scene categories. Then, by calculating the sum of scene confidence scores between scene category A and each scene category B, it is necessary to filter out all scene categories C from all scene categories B whose sum of scene confidence scores is greater than or equal to a preset mixed scene confidence threshold. After these two scene category filters, it can be assumed that the target image contains a mixed scene of scene category A and all scene categories C. At this point, image quality parameters matching the target image can be set based on the scene confidence scores of scene category A and all scene categories C.

[0113] As can be seen, this optional embodiment can not only perform image quality parameter tuning for specific scenes of the target image, but also for mixed scenes of the target image. This enriches the intelligent recognition method of scene categories of the target image by the image processing system, which is conducive to further improving the accuracy of image quality parameter tuning of the target image, thereby improving the effectiveness of image quality parameter tuning of the target image, and thus improving the image quality effect of the target image.

[0114] In yet another optional embodiment, after determining the first undetermined scene category from all the scene categories based on the scene confidence level corresponding to each scene category in the above steps, the method may further include:

[0115] Determine whether the first pending scene category is a special scene category. If it is determined that the first pending scene category is not a special scene category, trigger the above steps to analyze the comparison between the scene confidence corresponding to the first pending scene category and the preset scene confidence threshold group corresponding to the first pending scene category, and determine the first image quality parameter that matches the target image based on the comparison between the scene confidence corresponding to the first pending scene category and the preset scene confidence threshold group corresponding to the first pending scene category.

[0116] When the first pending scene category is determined to be a special scene category, it is determined whether the scene confidence corresponding to the first pending scene category is greater than or equal to the maximum scene confidence threshold corresponding to the first pending scene category. If so, the preset image quality parameter is determined as the first image quality parameter that matches the target image.

[0117] Specifically, this special scene category can be understood as any scene category other than those that the target recognition model can recognize.

[0118] As can be seen, this optional embodiment can determine the image quality parameters that match the target image for specific scene categories, enriching the intelligent determination method of image quality parameters by the image processing system, thereby improving the reliability and accuracy of image quality adjustment of the target image.

[0119] In another optional embodiment, the step of determining the first image quality parameter matching the target image based on the scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all third undetermined scene categories includes:

[0120] For the first undetermined scene category and all third undetermined scene categories, a scene feature extraction operation is performed on the target image to obtain the first scene feature corresponding to the first undetermined scene category and the second scene feature corresponding to each third undetermined scene category;

[0121] Based on the first scene features and all the second scene features, the scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories are adjusted to obtain the adjusted scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories.

[0122] Based on the adjusted scene confidence scores for the first undetermined scene category and the scene confidence scores for all third undetermined scene categories, determine the first image quality parameters that match the target image.

[0123] In this optional embodiment, for example, after determining the first undetermined scene category (scene category A) and all third undetermined scene categories (all scene categories C), scene features can be extracted from the target image for scene category A and all scene categories C. This scene feature extraction operation can be understood as further determining the scene size of scene category A and all scene categories C, and adjusting the scene confidence of scene category A and all scene categories C based on the above scene size. For example, if the scene size of scene category A accounts for 90% and the scene size of all scene categories C accounts for 10%, the scene confidence of scene category A can be adaptively increased, that is, the scene confidence of scene category A is mainly used as the basis for adjusting the image quality parameters of the target image.

[0124] As can be seen, this optional embodiment can adjust the scene confidence of a given scene category by extracting scene features, and can specifically adjust the image quality parameters of mixed scenes in the target image, thereby improving the effectiveness of image quality adjustment of mixed scenes and thus improving the image quality effect of mixed scenes.

[0125] Example 2

[0126] Please see Figure 2 , Figure 2 This is a flowchart illustrating an intelligent adjustment method for image quality parameters disclosed in an embodiment of the present invention. Figure 2 The described intelligent adjustment method for image quality parameters can be applied to adjusting the image quality parameters of smart display devices such as smart TVs and smartphones, and also to adjusting the image quality parameters of network service products such as live video and video conferencing. This invention does not limit the application of this method. Optionally, this method can be implemented by an image processing system, which can be integrated into an image processing device. This system can be a local server or a cloud server for managing the image processing (parameter adjustment) workflow, and this invention does not limit the application of this method. Figure 2 As shown, the intelligent adjustment method for image quality parameters can include the following operations:

[0127] 201. Using the target recognition model, perform image recognition operations on the target image to be recognized to obtain the image recognition result corresponding to the target image.

[0128] 202. Based on the image recognition results, determine the set of image information for the target image.

[0129] 203. Based on the scene information set, determine the first image quality parameters that match the target image.

[0130] 204. Determine whether the image information set of the target image also includes the target region information set of the target image; if the determination result of step 204 is no, trigger the execution of step 205; if the determination result of step 204 is yes, trigger the execution of step 206. Furthermore, step 204 can be executed synchronously with step 203, or step 204 can be executed after step 203 is completed.

[0131] In this embodiment of the invention, the target region information set includes a skin color region information set, or a skin color region information set and a face region information set. Each skin color region in the skin color region information set includes a first skin color confidence level for that skin color region, and each face region in the face region information set includes a second skin color confidence level for that face region. Specifically, determining whether the image information set of the target image also includes the target region information set of the target image can be understood as follows: the image recognition result obtained by the target recognition model may not necessarily include face recognition results and skin color recognition results. Therefore, for the determined image information set, it is also necessary to determine whether it also includes the target region information set. When the target region information set includes the face region information set, then the target region information set must also include the skin color region information set.

[0132] 205. Based on the first image quality parameters, perform a first adjustment operation on the current image quality parameters of the target image.

[0133] In this embodiment of the invention, for other descriptions of steps 201-203 and 205, please refer to the detailed description of steps 101-104 in Embodiment 1, and this embodiment of the invention will not repeat them.

[0134] 206. Based on the target region information set, determine the second image quality parameters that match the target image.

[0135] In this embodiment of the invention, specifically, based on the skin color region information set, or the skin color region information set and the face region information set, a second image quality parameter matching the target image can be determined. The second image quality parameter can be understood as a local region image quality parameter generated for the skin color region, or the skin color region and the face region, so as to adjust the local region image quality parameter of the target image.

[0136] 207. Based on the first image quality parameters and the second image quality parameters, perform a second adjustment operation on the current image quality parameters of the target image.

[0137] In this embodiment of the invention, specifically, performing a second adjustment operation on the current image quality parameters of the target image can be understood as adjusting the global region image quality parameters of the target image based on the first image quality parameters and further adjusting the local region image quality parameters of the target image based on the second image quality parameters. That is, adjusting the global scene region image quality parameters (such as contrast, brightness, sharpness, etc. of the scene region) of the target image based on the scene image quality parameters that match the confidence level corresponding to the scene category, and adjusting the skin region, or the skin region and face region image quality parameters (such as contrast, brightness, sharpness, etc. of non-face areas in the skin region, and contrast, brightness, sharpness, etc. of the face region) of the target image based on the skin region image quality parameters that match the skin region confidence level corresponding to the skin region, or the skin region and face region image quality parameters (such as contrast, brightness, sharpness, etc. of non-face areas in the skin region, and contrast, brightness, sharpness, etc. of the face region) of the target image based on the skin region image quality parameters that match the skin region confidence level corresponding to the skin region, or the skin region and face region image quality parameters (such as contrast, brightness, sharpness, etc. of non-face areas in the skin region, and contrast, brightness, sharpness, etc. of the face region) of the target image based on the skin region image quality parameters that match the skin region confidence level corresponding to the skin region, and the face region image quality parameters (such as contrast, brightness, sharpness, etc. of non-face areas in the skin region, and contrast, brightness, sharpness, etc. of the face region) of the target image based on the skin region image quality parameters that match the skin region confidence level corresponding to the face region.

[0138] As can be seen, implementing the embodiments of the present invention can not only adjust the image quality parameters of the global region of the target image through the confidence of the scene category, but also adjust the image quality parameters of the local region of the target image through the skin color confidence of the face region and skin color region. This enriches the intelligent adjustment method of the image processing system for the image quality parameters of the target image. This is beneficial to optimize the image quality of the local region of the target image, thereby improving the overall display effect and visual effect of the target image, and thus enhancing the viewer's visual experience of the target image.

[0139] In an optional embodiment, step 206 above, determining the second image quality parameters matching the target image based on the target region information set, includes:

[0140] When the target region information set only includes the skin color region information set, based on the first skin color confidence of all skin color regions, determine all skin color regions whose first skin color confidence is greater than or equal to the preset first confidence threshold, and use them as all first target skin color regions.

[0141] Based on the first skin color confidence scores of all first target skin color regions, determine the second image quality parameters that match the target image;

[0142] When the target region information set includes a skin color region information set and a face region information set, the second image quality parameters that match the target image are determined based on the first skin color confidence score of all skin color regions and the second skin color confidence score of all face regions.

[0143] In this optional embodiment, for example, determining all skin color regions with a skin color confidence level greater than or equal to a preset first confidence threshold from all skin color regions can be understood as retaining skin color regions with a confidence level greater than or equal to the preset first confidence threshold in the target image, while ignoring the image quality parameters of these non-target skin color regions with a confidence level less than the preset first confidence threshold. That is, it is not necessary to perform image quality parameter tuning on these non-target skin color regions, or image quality parameter tuning can be performed on these non-target skin color regions using default image quality parameters.

[0144] As can be seen, this optional embodiment can perform local image quality parameter adjustment on the skin color area of ​​the target image in a targeted and evidence-based manner, which is beneficial to improving the reliability and accuracy of local image quality parameter adjustment of the target image, thereby improving the effectiveness of local image quality parameter adjustment of the target image, and thus improving the local and global image quality of the target image.

[0145] In another optional embodiment, the step of determining the second image quality parameters matching the target image based on the first skin color confidence scores of all skin color regions and the second skin color confidence scores of all face regions includes:

[0146] Based on all face regions, identify all first regions to be processed that overlap with the corresponding face regions from all skin color regions, and identify all second regions to be processed that do not overlap with the corresponding face regions from all skin color regions.

[0147] Based on the location information of each first region to be processed, determine the first weight of each first region to be processed and the second weight of each second region to be processed;

[0148] Based on the first weight of each first region to be processed, an adjustment operation is performed on the second skin color confidence of the face region corresponding to each first region to be processed to update the second skin color confidence of all face regions; and based on the second weight of each second region to be processed, an adjustment operation is performed on the first skin color confidence of the skin color region corresponding to each second region to be processed to update the first skin color confidence of all skin color regions.

[0149] After updating the first skin confidence of all skin color regions and the second skin confidence of all face regions, based on the first skin confidence of all skin color regions, all skin color regions whose first skin confidence is greater than or equal to a preset first confidence threshold are identified from all skin color regions as all second target skin color regions. Similarly, based on the second skin confidence of all face regions, all face regions whose second skin confidence is greater than or equal to a preset second confidence threshold are identified from all face regions as all target face regions.

[0150] Based on the first skin color confidence scores of all second target skin color regions and the second skin color confidence scores of all target face regions, determine the second image quality parameters that match the target image.

[0151] In this optional embodiment, the location information of each first region to be processed includes the distance between the first region to be processed and all second regions to be processed. Optionally, the first confidence threshold and the second confidence threshold can be the same or different; that is, the evaluation criteria for the skin color confidence of skin color regions and the skin color confidence of face regions can be the same or different. Specifically, determining all first regions to be processed that overlap with the corresponding face regions from all skin color regions, and determining all second regions to be processed that do not overlap with the corresponding face regions from all skin color regions, can be understood as follows: In actual image processing, when there are face regions in the image, the face regions will inevitably overlap with the skin color regions. This overlapping region can be used as the basis for adjusting the parameters of local regions (face and skin color regions), thereby selectively adjusting the image quality of face and non-face skin color regions.

[0152] Further, in this optional embodiment, determining a first weight for each first region to be processed and a second weight for each second region to be processed based on the location information of each first region to be processed includes:

[0153] For each second region to be processed, the weights of all pixels in the second region to be processed are determined based on the distance between the second region to be processed and all the first regions to be processed, and these weights are used as the second weights of the second region to be processed.

[0154] Based on the second weights of all the second regions to be processed, determine the first weight of each first region to be processed.

[0155] Specifically, the first weight of each first region to be processed can be used to represent the weight of all pixels in each first region to be processed. Further, based on the first weight of each first region to be processed, performing an adjustment operation on the second skin color confidence of the face region corresponding to each first region to be processed can include: performing an adjustment operation on the skin color confidence of each pixel in the face region corresponding to each first region to be processed based on the weight of each pixel in each first region to be processed. Further still, based on the second weight of each second region to be processed, performing an adjustment operation on the first skin color confidence of the skin color region corresponding to each second region to be processed can include: performing an adjustment operation on the skin color confidence of each pixel in the skin color region corresponding to each second region to be processed based on the weight of each pixel in each second region to be processed.

[0156] For example, in a target image containing both face and skin regions, the weight of non-face regions (i.e., the second region to be processed) within the skin region can be determined based on their distance from the face region. If non-face region A within the skin region is closer to the face region, its weight can be adaptively increased; conversely, if non-face region B within the skin region is farther from the face region, its weight can be adaptively decreased. Based on the weights of non-face regions A and B, the weight of the face region is determined. Then, based on these weights, the skin confidence scores of non-face regions A and B, as well as the face region, are adjusted accordingly. In other words, the focus of parameter adjustment differs for non-face regions and the skin region within the face region, ensuring that the image quality of the face region is highlighted and reducing the error in the skin confidence scores of non-face regions within the determined skin region.

[0157] As can be seen, this optional embodiment can adjust the skin color region information through the face region information, which reflects the intelligent way the image processing system adjusts the image quality parameters of the face region and the non-face region in the skin color region. This is conducive to further improving the accuracy of image quality parameter adjustment of the face region and the non-face region in the skin color region of the target image, thereby highlighting the image quality effect of the face region in the target image and reducing the error of the skin color confidence of the non-face region in the determined skin color region, thus improving the user's visual experience of the target image.

[0158] Example 3

[0159] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of an intelligent image quality parameter adjustment device disclosed in an embodiment of the present invention. Figure 3 As shown, the intelligent adjustment device for image quality parameters may include:

[0160] Image recognition module 301 is used to perform image recognition operation on the target image to be recognized through the target recognition model, and obtain the image recognition result corresponding to the target image;

[0161] The first determining module 302 is used to determine the image information set of the target image based on the image recognition result;

[0162] The second determining module 303 is used to determine the first image quality parameters that match the target image based on the scene information set;

[0163] The adjustment module 304 is used to perform a first adjustment operation on the current image quality parameters of the target image according to the first image quality parameters.

[0164] In this embodiment of the invention, the target recognition model includes a scene recognition model, a face detection model, and a skin color detection model; the image information set of the target image includes a scene information set of the target image, and each scene information in the scene information set includes the scene confidence level corresponding to each scene category.

[0165] It is evident that implementation Figure 3 The intelligent image quality parameter adjustment device described herein can perform global image quality parameter adjustment based on the directly obtained scene category confidence score after the image is identified by the recognition model. It can also perform local image quality parameter adjustment based on the directly obtained skin color confidence score of face and non-face regions. It can obtain scene category confidence score without matching scene features with a real scene category library. In this way, it not only solves the problem of limited image scene category recognition in traditional image quality parameter adjustment methods, but also improves the accuracy of image quality parameter adjustment of the whole image by adjusting the image quality of local areas (face and skin color regions), thereby enhancing the overall image quality effect and improving the user's visual experience.

[0166] In an optional embodiment, the second determining module 303 includes:

[0167] The first determining submodule 3031 is used to determine the first undetermined scene category from all scene categories based on the scene confidence level corresponding to each scene category;

[0168] The analysis submodule 3032 is used to analyze the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence level threshold group corresponding to the first undetermined scene category;

[0169] The second determining submodule 3033 is used to determine the first image quality parameters that match the target image based on the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence threshold group corresponding to the first undetermined scene category.

[0170] In this optional embodiment, the first undetermined scene category is the scene category with the highest scene confidence among all scene categories; each scene category has a corresponding preset scene confidence threshold group, and the preset scene confidence threshold group corresponding to each scene category includes the maximum scene confidence threshold and the minimum scene confidence threshold corresponding to that scene category.

[0171] It is evident that implementation Figure 4 The intelligent image quality parameter adjustment device described herein can determine the scene category present in the target image by comparing the confidence level corresponding to the scene category with its own preset scene confidence threshold group. At the same time, it can adjust the image quality parameters of the target image by using the first undetermined scene category with the maximum scene confidence. This helps to reduce the impact of limited scene category recognition, expands the range of scene categories that the image processing system can recognize, and thus improves the reliability and accuracy of the overall target image quality parameter adjustment, thereby improving the image quality effect of the target image and the viewer's visual experience.

[0172] In another alternative embodiment, the second determining submodule 3033 includes:

[0173] The determining unit 30331 is configured to: determine a preset image quality parameter as a first image quality parameter matching the target image when the scene confidence corresponding to the first undetermined scene category is less than the minimum scene confidence threshold corresponding to the first undetermined scene category; determine a first image quality parameter matching the target image based on the scene confidence corresponding to the first undetermined scene category when the scene confidence corresponding to the first undetermined scene category is greater than or equal to the maximum scene confidence threshold corresponding to the first undetermined scene category; and determine all second undetermined scene categories other than the first undetermined scene category from all scene categories when the scene confidence corresponding to the first undetermined scene category is greater than or equal to the minimum scene confidence threshold corresponding to the first undetermined scene category and less than the maximum scene confidence threshold corresponding to the first undetermined scene category.

[0174] The calculation unit 30332 is used to calculate the sum of the scene confidence scores corresponding to the first undetermined scene category and the scene confidence scores corresponding to each of the second undetermined scene categories;

[0175] The determining unit 30331 is further configured to determine all second undetermined scene categories from all second undetermined scene categories whose sum of scene confidence is greater than or equal to a preset mixed scene confidence threshold, as all third undetermined scene categories; and determine the first image quality parameter that matches the target image based on the scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all third undetermined scene categories.

[0176] In this optional embodiment, the scene confidence level corresponding to each second undetermined scene category is greater than or equal to the minimum scene confidence level threshold corresponding to that second undetermined scene category.

[0177] It is evident that implementation Figure 4 The intelligent image quality parameter adjustment device described can not only adjust the image quality parameters for specific scenes of the target image, but also for mixed scenes of the target image. This enriches the intelligent recognition method of the target image scene category by the image processing system, which is conducive to further improving the accuracy of the image quality parameter adjustment of the target image, thereby improving the effectiveness of the image quality parameter adjustment of the target image, and thus improving the image quality effect of the target image.

[0178] In another optional embodiment, the determining unit 30331 determines the first image quality parameter matching the target image based on the scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all third undetermined scene categories in the following specific manner:

[0179] For the first undetermined scene category and all third undetermined scene categories, a scene feature extraction operation is performed on the target image to obtain the first scene feature corresponding to the first undetermined scene category and the second scene feature corresponding to each third undetermined scene category;

[0180] Based on the first scene features and all the second scene features, the scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories are adjusted to obtain the adjusted scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories.

[0181] Based on the adjusted scene confidence scores for the first undetermined scene category and the scene confidence scores for all third undetermined scene categories, determine the first image quality parameters that match the target image.

[0182] It is evident that implementation Figure 4 The intelligent adjustment device for image quality parameters described herein can adjust the scene confidence of a given scene category based on the extracted scene features. It can specifically adjust the image quality parameters of mixed scenes in the target image, thereby improving the effectiveness of image quality adjustment for mixed scenes and thus enhancing the image quality effect of mixed scenes.

[0183] In yet another alternative embodiment, the device may further include:

[0184] The judgment module 305 is used to determine whether the image information set of the target image still includes the target region information set of the target image before the adjustment module 304 performs the first adjustment operation on the current image quality parameters of the target image according to the first image quality parameters; when it is determined that the image information set of the target image does not include the target region information set of the target image, the adjustment module 304 is triggered to perform the first adjustment operation on the current image quality parameters of the target image according to the first image quality parameters.

[0185] The third determining module 306 is used to determine the second image quality parameter that matches the target image based on the target region information set when the determining module 305 determines that the image information set of the target image also includes the target region information set of the target image.

[0186] The adjustment module 304 is also used to perform a second adjustment operation on the current image quality parameters of the target image based on the first image quality parameters and the second image quality parameters.

[0187] In this optional embodiment, the target region information set includes a skin color region information set, or a skin color region information set and a face region information set, wherein each skin color region information in the skin color region information set includes a first skin color confidence level for each skin color region, and each face region information in the face region information set includes a second skin color confidence level for each face region.

[0188] It is evident that implementation Figure 4 The intelligent image quality parameter adjustment device described can not only adjust the image quality parameters of the global area of ​​the target image based on the confidence level of the scene category, but also adjust the image quality parameters of the local area of ​​the target image based on the skin color confidence level of the face area and skin color area. This enriches the intelligent adjustment method of the image processing system for the image quality parameters of the target image. This is beneficial to optimize the image quality of the local area of ​​the target image, thereby improving the overall display effect and visual effect of the target image, and thus enhancing the viewer's visual experience of the target image.

[0189] In yet another optional embodiment, the third determining module 306 includes:

[0190] The third determining submodule 3061 is used to determine, when the target region information set only includes the skin color region information set, all skin color regions whose first skin color confidence is greater than or equal to a preset first confidence threshold, as all first target skin color regions, based on the first skin color confidence of all skin color regions; and to determine a second image quality parameter that matches the target image based on the first skin color confidence of all first target skin color regions. When the target region information set includes both the skin color region information set and the face region information set, the second image quality parameter that matches the target image is determined based on the first skin color confidence of all skin color regions and the second skin color confidence of all face regions.

[0191] It is evident that implementation Figure 4 The intelligent image quality parameter adjustment device described herein can perform local image quality parameter adjustment on the skin color area of ​​the target image in a targeted and evidence-based manner, which helps to improve the reliability and accuracy of local image quality parameter adjustment of the target image, thereby improving the effectiveness of local image quality parameter adjustment of the target image, and thus helping to improve the local and global image quality effect of the target image.

[0192] In another optional embodiment, the third determining submodule 3061 determines the second image quality parameters matching the target image based on the first skin color confidence scores of all skin color regions and the second skin color confidence scores of all face regions in the following specific manner:

[0193] Based on all face regions, identify all first regions to be processed that overlap with the corresponding face regions from all skin color regions, and identify all second regions to be processed that do not overlap with the corresponding face regions from all skin color regions.

[0194] Based on the location information of each first region to be processed, a first weight for each first region to be processed and a second weight for each second region to be processed are determined; the location information of each first region to be processed includes the distance between the first region to be processed and all second regions to be processed.

[0195] Based on the first weight of each first region to be processed, an adjustment operation is performed on the second skin color confidence of the face region corresponding to each first region to be processed to update the second skin color confidence of all face regions; and based on the second weight of each second region to be processed, an adjustment operation is performed on the first skin color confidence of the skin color region corresponding to each second region to be processed to update the first skin color confidence of all skin color regions.

[0196] After updating the first skin confidence of all skin color regions and the second skin confidence of all face regions, based on the first skin confidence of all skin color regions, all skin color regions whose first skin confidence is greater than or equal to a preset first confidence threshold are identified from all skin color regions as all second target skin color regions. Similarly, based on the second skin confidence of all face regions, all face regions whose second skin confidence is greater than or equal to a preset second confidence threshold are identified from all face regions as all target face regions.

[0197] Based on the first skin color confidence scores of all second target skin color regions and the second skin color confidence scores of all target face regions, determine the second image quality parameters that match the target image.

[0198] It is evident that implementation Figure 4 The described intelligent image quality parameter adjustment device can adjust skin color region information based on face region information, reflecting the intelligent way the image processing system adjusts the image quality parameters of non-face regions in face and skin color regions. This helps to further improve the accuracy of image quality parameter adjustment in face and non-face regions in the target image, thereby highlighting the image quality effect of face regions in the target image and reducing the error of skin color confidence in non-face regions in the determined skin color region, thus improving the user's visual experience of the target image.

[0199] Example 4

[0200] Please see Figure 5 , Figure 5 This is a schematic diagram of the structure of another intelligent image quality parameter adjustment device disclosed in an embodiment of the present invention. For example... Figure 5 As shown, the intelligent adjustment device for image quality parameters may include:

[0201] Memory 401 storing executable program code;

[0202] Processor 402 coupled to memory 401;

[0203] The processor 402 calls the executable program code stored in the memory 401 to execute the steps in the intelligent adjustment method for image quality parameters described in Embodiment 1 or Embodiment 2 of the present invention.

[0204] Example 5

[0205] This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute the steps in the intelligent adjustment method for image quality parameters described in Embodiment 1 or Embodiment 2 of this invention.

[0206] Example 6

[0207] This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the intelligent adjustment method for image quality parameters described in Embodiment 1 or Embodiment 2.

[0208] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0209] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0210] Finally, it should be noted that the intelligent adjustment method and apparatus for image quality parameters disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent adjustment of image quality parameters, characterized in that, The method includes: The target recognition model performs image recognition operations on the target image to be recognized, and obtains the image recognition result corresponding to the target image; the target recognition model includes a scene recognition model, a face detection model, and a skin color detection model; Based on the image recognition results, an image information set for the target image is determined; the image information set for the target image includes a scene information set for the target image, and each scene information in the scene information set includes a scene confidence score corresponding to each scene category; Based on the scene information set, determine the first image quality parameters that match the target image; Determine whether the image information set of the target image also includes the target region information set of the target image; the target region information set includes a skin color region information set, or the skin color region information set and a face region information set, wherein each skin color region information in the skin color region information set includes a first skin color confidence level for each skin color region, and each face region information in the face region information set includes a second skin color confidence level for each face region; When it is determined that the image information set of the target image also includes the target region information set of the target image, a second image quality parameter matching the target image is determined according to the target region information set, and a second adjustment operation is performed on the current image quality parameter of the target image according to the first image quality parameter and the second image quality parameter. The step of determining the second image quality parameters matching the target image based on the target region information set includes: When the target region information set includes the skin color region information set and the face region information set, based on all the face regions, all first regions to be processed that overlap with the corresponding face regions and all second regions to be processed that do not overlap with the corresponding face regions are determined from all the skin color regions. Based on the distance between each first region to be processed and all second regions to be processed, a first weight and a second weight of each first region to be processed are determined. Based on the first weight of each first region to be processed, the second skin color confidence of the face region corresponding to each first region to be processed is adjusted. Based on the second weight of each second region to be processed, the first skin color confidence of the skin color region corresponding to each second region to be processed is adjusted. This is to update the first skin color confidence of all skin color regions and the second skin color confidence of all face regions. After the update is complete, based on the first skin color confidence score of all the skin color regions and the second skin color confidence score of all the face regions, a second image quality parameter matching the target image is determined.

2. The intelligent adjustment method for image quality parameters according to claim 1, characterized in that, Each of the aforementioned scene categories has a corresponding preset scene confidence threshold group. The preset scene confidence threshold group corresponding to each scene category includes the maximum scene confidence threshold and the minimum scene confidence threshold corresponding to that scene category. The step of determining the first image quality parameter matching the target image based on the scene information set includes: Based on the scene confidence level corresponding to each scene category, a first undetermined scene category is determined from all the scene categories; the first undetermined scene category is the scene category with the highest scene confidence level among all the scene categories; Analyze the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence threshold group corresponding to the first undetermined scene category; Based on the comparison between the scene confidence level corresponding to the first undetermined scene category and the preset scene confidence threshold group corresponding to the first undetermined scene category, a first image quality parameter matching the target image is determined.

3. The intelligent adjustment method for image quality parameters according to claim 2, characterized in that, The step of determining the first image quality parameter matching the target image based on the comparison between the scene confidence score corresponding to the first undetermined scene category and the preset scene confidence score threshold group corresponding to the first undetermined scene category includes: When the scene confidence level corresponding to the first undetermined scene category is less than the minimum scene confidence threshold corresponding to the first undetermined scene category, the preset image quality parameter is determined as the first image quality parameter that matches the target image. When the scene confidence level corresponding to the first undetermined scene category is greater than or equal to the maximum scene confidence level threshold corresponding to the first undetermined scene category, the first image quality parameter matching the target image is determined based on the scene confidence level corresponding to the first undetermined scene category. When the scene confidence level corresponding to the first undetermined scene category is greater than or equal to the minimum scene confidence level threshold corresponding to the first undetermined scene category and less than the maximum scene confidence level threshold corresponding to the first undetermined scene category, all second undetermined scene categories other than the first undetermined scene category are determined from all the scene categories; the scene confidence level corresponding to each second undetermined scene category is greater than or equal to the minimum scene confidence level corresponding to that second undetermined scene category; Calculate the sum of the scene confidence scores corresponding to the first undetermined scene category and the scene confidence scores corresponding to each of the second undetermined scene categories; From all the second undetermined scenario categories, determine all the second undetermined scenario categories whose sum of scenario confidence is greater than or equal to a preset mixed scenario confidence threshold, and use them as all third undetermined scenario categories; Based on the scene confidence scores corresponding to the first undetermined scene category and all the scene confidence scores corresponding to the third undetermined scene categories, a first image quality parameter matching the target image is determined.

4. The intelligent adjustment method for image quality parameters according to claim 3, characterized in that, The step of determining the first image quality parameter matching the target image based on the scene confidence score corresponding to the first undetermined scene category and the scene confidence scores corresponding to all the third undetermined scene categories includes: For the first undetermined scene category and all the third undetermined scene categories, a scene feature extraction operation is performed on the target image to obtain the first scene feature corresponding to the first undetermined scene category and the second scene feature corresponding to each of the third undetermined scene categories; Based on the first scene features and all the second scene features, an adjustment operation is performed on the scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories to obtain the adjusted scene confidence corresponding to the first undetermined scene category and the scene confidence corresponding to all the third undetermined scene categories. Based on the adjusted scene confidence scores corresponding to the first undetermined scene category and the scene confidence scores corresponding to all the third undetermined scene categories, a first image quality parameter matching the target image is determined.

5. The intelligent adjustment method for image quality parameters according to any one of claims 1-4, characterized in that, The method further includes: When it is determined that the image information set of the target image does not include the target region information set of the target image, a first adjustment operation is performed on the current image quality parameters of the target image according to the first image quality parameters.

6. The intelligent adjustment method for image quality parameters according to any one of claims 1-4, characterized in that, The step of determining the second image quality parameters matching the target image based on the target region information set further includes: When the target region information set only includes the skin color region information set, based on the first skin color confidence of all the skin color regions, all skin color regions whose first skin color confidence is greater than or equal to a preset first confidence threshold are determined from all the skin color regions and are taken as all first target skin color regions. Based on the first skin color confidence of all the first target skin color regions, a second image quality parameter matching the target image is determined.

7. The intelligent adjustment method for image quality parameters according to any one of claims 1-4, characterized in that, The step of determining the second image quality parameters matching the target image based on the first skin color confidence score of all the skin color regions and the second skin color confidence score of all the face regions includes: Based on the first skin confidence of all skin color regions, all skin color regions whose first skin confidence is greater than or equal to a preset first confidence threshold are determined from all skin color regions as all second target skin color regions; and based on the second skin confidence of all face regions, all face regions whose second skin confidence is greater than or equal to a preset second confidence threshold are determined from all face regions as all target face regions. Based on the first skin color confidence scores of all second target skin color regions and the second skin color confidence scores of all target face regions, a second image quality parameter matching the target image is determined.

8. An intelligent adjustment device for image quality parameters, characterized in that, The device is used to perform the intelligent adjustment method for image quality parameters as described in any one of claims 1-7, and the device comprises: The image recognition module is used to perform image recognition operations on the target image to be recognized through a target recognition model, and obtain the image recognition result corresponding to the target image; the target recognition model includes a scene recognition model, a face detection model, and a skin color detection model. The first determining module is used to determine the image information set of the target image based on the image recognition result; the image information set of the target image includes the scene information set of the target image, and each scene information in the scene information set includes the scene confidence level corresponding to each scene category; The second determining module is used to determine a first image quality parameter that matches the target image based on the scene information set; The adjustment module is used to perform a first adjustment operation on the current image quality parameters of the target image based on the first image quality parameters.

9. An intelligent adjustment device for image quality parameters, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the intelligent adjustment method for image quality parameters as described in any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, are used to execute the intelligent adjustment method for image quality parameters as described in any one of claims 1-7.

Images