AI image processing system and method

Abstract

The invention discloses an AI image processing system and method, and relates to the technical field of AI image processing, and the system comprises an image collection module, an image feature extraction module, an image processing mode detection module, an image processing mode recommendation module, an image processing parameter recommendation module, and an image processing module. The image collection module is used for acquiring images, selecting and storing unprocessed images to obtain original images, and selecting and storing processed images corresponding to the original images; according to the AI image processing system and method, the processing features of the processed image can be found and recognized through the training model, so that the processing mode of the processed image can be quickly obtained, and the model can be trained by using the processing mode subsequently; and the relationship between the processing mode and the processing parameter adopted by the processing image with a good display effect and the original image is found through the training model, so that the model can obtain the recommended processing mode and the processing parameter through the original image, the image is quickly processed, and the display effect of the image is improved.

Description

Technical Field

[0001] The present invention relates to the field of AI image processing technology, and in particular to an AI image processing system and method. Background Art

[0002] Image processing refers to the technology of using computers to analyze images to achieve the desired results. It generally includes three parts: image compression, enhancement and restoration, matching, description and recognition. In order for the image to be presented to people with better display and viewing effects, it is generally necessary to process the image.

[0003] The prior art of publication number CN117611578A discloses an image processing method and an image processing system, and relates to the field of image processing technology. When the system is running, image acquisition is performed through an image acquisition module to obtain original image data, feature extraction is performed through an extraction module to obtain brightness information and pixel information in the original image, the brightness information and pixel information are preprocessed through a preprocessing module to form a first data set and a second data set, and calculation is performed through an image calculation module to obtain: a visual anomaly index Sjzs, which is matched with a preset visual anomaly threshold S through an evaluation module to obtain a visual anomaly evaluation strategy scheme, and finally the content of the visual anomaly evaluation strategy scheme is specifically executed and notified through an execution module, thereby detecting visual problems of light spots and shadows appearing in the captured image, and providing personalized adjustment suggestions, so as to achieve the purpose of optimizing image quality, improving work efficiency, and reducing the number of device shots.

[0004] However, in the process of image processing, how to determine the image processing method and processing parameters in order to automatically process the image to achieve better display effects, reduce the difficulty of image processing, and simplify the steps of image processing is a problem that needs to be solved. Summary of the invention

[0005] The purpose of the present invention is to provide an AI image processing system and method to solve the above-mentioned deficiencies in the prior art.

[0006] In order to achieve the above-mentioned object, the present invention provides the following technical solutions: an AI image processing system, comprising an image collection module, an image feature extraction module, an image processing method detection module, an image processing method recommendation module, an image processing parameter recommendation module, and an image processing module;

[0007] The image collection module is used to acquire images, select unprocessed images to save, obtain original images, and select processed images corresponding to the original images to save, wherein when selecting processed images, manually select processed images with better display effects to save corresponding to the original images;

[0008] The image feature extraction module is used to extract features from the original image and the processed image, and can select pre-trained edge detection algorithms, VGGNet, ResNet, Inception, SIFT, SURF, ORB, etc.;

[0009] The image processing method detection module is used to select an original image as a template image, and set a plurality of image processing methods, respectively use the image processing methods to process the template image, obtain a corresponding template processing image, analyze the difference between the template processing image and the corresponding template image, obtain the difference characteristics of each image processing method, train a first machine learning model based on the template processing image, the difference characteristics, and the image processing method, obtain a processing method detection model, the processing method detection model is used to identify the difference characteristics in the input processing image and the pixel coordinates of the difference characteristics in the processing image, obtain the difference coordinates, and output the corresponding image processing method and the difference coordinates;

[0010] The image processing method recommendation module is used to obtain the image features of the original image corresponding to the processed image at the difference coordinate position of the processed image, obtain the processing position feature of the original image, train the second machine learning model based on the original image, the image processing method and the processing position feature, and obtain the processing method recommendation model. The processing method recommendation model is used to obtain the recommended image processing method based on the input original image, detect the processing position feature, and output the processing position coordinates. The processing position coordinates are the pixel coordinates of the processing position feature, and its pixel coordinates in the original image can be directly obtained through the detected processing position feature;

[0011] The image processing parameter recommendation module is used to analyze the difference between the processing position feature of each original image and the difference feature of the corresponding processed image to obtain the difference feature parameter, and train the third machine learning model based on the original image, the image processing method and the difference feature parameter to obtain the processing parameter prediction model, and the processing parameter prediction model is used to output the predicted difference feature parameter based on the input original image and the image processing method;

[0012] The image processing module is used to obtain a new original image, input the new original image into the processing method recommendation model to obtain the corresponding image processing method and processing position coordinates, and then input the new original image and the obtained image processing method into the processing parameter prediction model to obtain corresponding difference feature parameters, and process the new original image based on the obtained image processing method, processing position coordinates and difference feature parameters to obtain a corresponding new processed image.

[0013] Furthermore, the image processing method detection module is used to select an original image as a template image, and set a plurality of image processing methods, respectively use the image processing methods to process the template image, obtain a corresponding template-processed image, analyze the difference between the template-processed image and the corresponding template image, obtain the difference characteristics of each image processing method, and train a first machine learning model based on the template-processed image, the difference characteristics, and the image processing method to obtain a processing method detection model, including the following steps:

[0014] Select the original image as the template image and set various image processing methods;

[0015] Using each image processing method to process the template image respectively to obtain a corresponding template processed image;

[0016] Based on the template image and the template-processed image, a processed difference image of the template image and the template-processed image is obtained, wherein a difference image can be generated by subtracting the template-processed image from the template image, an area with a zero pixel value indicates that the two images are the same, and a non-zero area indicates a processed area, and an image of the processed area in the template-processed image is selected as the processed difference image;

[0017] grayscale the processed difference image, and use the image feature extraction module to process the processed difference image after the grayscale processing to obtain corresponding difference features;

[0018] Setting a plurality of processing method labels to correspond to image processing methods one by one, and associating the difference features with the corresponding processing method labels;

[0019] Based on the difference features, template-processed images and image processing labels, a first machine learning model is trained to obtain a processing method detection model. During training, the template-processed image is used as input, and the difference features and image processing labels are used as outputs, wherein the difference features are the output of the hidden layer, and the processing labels are the output of the output layer. The processing method detection model is used to detect the difference features in the processed image based on the input processed image, obtain the pixel coordinates of the difference features in the processed image to obtain the difference coordinates, and output the processing method label associated with the difference features and the difference coordinates.

[0020] The present invention is not limited to specific machine learning models, such as convolutional neural networks (CNN), support vector machines (SVM), random forests (Random Forest) generative adversarial networks (GANs), etc.

[0021] Furthermore, the system also includes a processing method verification module, which is used for the user to select the user to verify that the relationship is correct, and the processed image and the corresponding processing method label and pixel coordinates are saved as samples.

[0022] Furthermore, the processing method recommendation module is also used to use the samples to train and update the processing method detection model when the samples reach a set first sample quantity.

[0023] Furthermore, the image processing method recommendation module trains a second machine learning model based on the original image, the image processing method and the processing position features to obtain a processing method recommendation model, wherein the processing method recommendation model is used to obtain a recommended image processing method based on the input original image, and comprises the following steps:

[0024] Select an original image for which a corresponding processed image exists;

[0025] Obtaining image processing methods of an original image and a corresponding processed image, and obtaining a processing method label corresponding to the image processing method;

[0026] Input the original image into the image feature extraction module to obtain the corresponding original image features;

[0027] Based on the original image features, the processing method labels and the processing position features, a second machine learning model is trained to obtain a processing method recommendation model. During training, the original image features are used as input, and the processing position features and the processing method labels are used as output, wherein the processing position features are the hidden layer outputs, and the processing method labels are the output layer outputs. The model is used to extract the original image features from the input original image through an image feature extraction module, retrieve the corresponding processing position features based on the original image features, obtain the pixel coordinates of the processing position features to obtain the processing position coordinates, and output the corresponding processing method labels and the processing position coordinates.

[0028] Furthermore, the image processing parameter recommendation module trains a third machine learning model based on the original image, the image processing method and the difference feature parameters to obtain a processing parameter prediction model, including the following steps:

[0029] Input the original image into the image feature extraction module to obtain the corresponding original image features;

[0030] Classifying the original image and the corresponding processed image according to different processing methods to obtain multiple processing groups;

[0031] Extracting the processing position features and the difference features corresponding to the processing methods corresponding to the processing groups in each processing group, respectively, and obtaining difference feature parameters of the difference features and the corresponding difference position features;

[0032] Based on the original image features and difference feature parameters of each processing group, a third machine learning model is trained to obtain a processing parameter prediction sub-model, wherein the original image features are used as input and the difference feature parameters are used as output during training;

[0033] The processing parameter prediction sub-models of all processing groups are merged to obtain a processing parameter prediction model. The processing parameter prediction model selects the corresponding processing parameter prediction sub-model according to the input processing method label, extracts the original image features through the image feature extraction module, and outputs the corresponding difference feature parameters based on the original image features.

[0034] An AI image processing method comprises the following steps:

[0035] S1, select the original image and the corresponding processed image through the image collection module and save them for later use;

[0036] S2, extracting features from the original image and the processed image through an image feature extraction module;

[0037] S3, training the first machine learning model through the image processing method detection module to obtain a processing method detection model;

[0038] S4, training the second machine learning model through the image processing method recommendation module to obtain a processing method recommendation model;

[0039] S5. Train the third machine learning model through the image processing parameter recommendation module to obtain a processing parameter prediction model;

[0040] S6. A new original image to be processed is inputted through the image collection module, and the image processing module processes the new original image to obtain a new processed image through a processing method recommendation model and a processing parameter prediction model.

[0041] Compared with the prior art, the present invention provides an AI image processing system and method, which can discover and identify the processing features of a processed image through a training model by setting an image collection module, an image feature extraction module, an image processing method detection module, an image processing method recommendation module, an image processing parameter recommendation module, and an image processing module, so as to quickly obtain the processing method of the processed image, which is convenient for subsequent use of the processing method to train the model; and discover the relationship between the processed image with good display effect and the processing method and processing parameters adopted by the original image through the training model, so that the model can obtain the recommended processing method and processing parameters through the original image, quickly process the image, and improve the display effect of the image. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0043] Figure 1A system structure block diagram provided for an embodiment of the present invention;

[0044] Figure 2 A diagram of the method steps provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0046] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0047] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0048] Example embodiments will be described more fully below with reference to the accompanying drawings, but the example embodiments may be embodied in different forms and should not be construed as limited to the embodiments set forth herein. On the contrary, the purpose of providing these embodiments is to make the present disclosure thorough and complete and to enable those skilled in the art to fully understand the scope of the present disclosure.

[0049] In the absence of conflict, the various embodiments of the present disclosure and the various features therein may be combined with each other.

[0050] As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0051] The terms used herein are only used to describe specific embodiments and are not intended to limit the present disclosure. As used herein, the singular forms "a", "an" and "the" are also intended to include the plural forms, unless the context clearly indicates otherwise. It will also be understood that when the terms "comprising" and / or "made of" are used in this specification, the presence of the features, wholes, steps, operations, elements and / or components is specified, but the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof is not excluded.

[0052] The embodiments described herein may be described with reference to plan views and / or cross-sectional views by means of idealized schematic diagrams of the present disclosure. Therefore, the example illustrations may be modified according to manufacturing techniques and / or tolerances. Therefore, the embodiments are not limited to the embodiments shown in the accompanying drawings, but include modifications of the configurations formed based on the manufacturing process. Therefore, the regions illustrated in the accompanying drawings have schematic properties, and the shapes of the regions shown in the figures illustrate the specific shapes of the regions of the elements, but are not intended to be limiting.

[0053] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and the present disclosure, and will not be interpreted as having an idealized or overly formal meaning unless explicitly defined as such herein.

[0054] See also Figure 1 , an AI image processing system, comprising an image collection module, an image feature extraction module, an image processing method detection module, an image processing method recommendation module, an image processing parameter recommendation module, and an image processing module;

[0055] The image collection module is used to obtain images, select unprocessed images to save, obtain original images, and select processed images corresponding to the original images to save, wherein when selecting processed images, manually select processed images with better display effects to save corresponding to the original images;

[0056] The image feature extraction module is used to extract features from the original image and the processed image. You can choose pre-trained edge detection algorithms, VGGNet, ResNet, Inception, SIFT, SURF, ORB, etc.

[0057] The image processing method detection module is used to select the original image as the template image, and set a plurality of image processing methods, respectively use the image processing methods to process the template image, obtain the corresponding template processing image, analyze the difference between the template processing image and the corresponding template image, obtain the difference characteristics of each image processing method, and train the first machine learning model based on the template processing image, the difference characteristics, and the image processing method to obtain the processing method detection model. The processing method detection model is used to identify the difference characteristics in the input processing image and the pixel coordinates of the difference characteristics in the processing image, obtain the difference coordinates, and output the corresponding image processing method and the difference coordinates, including the following steps:

[0058] (1) Select the original image as the template image and set multiple image processing methods;

[0059] (2) using each image processing method to process the template image respectively to obtain a corresponding template-processed image;

[0060] (3) based on the template image and the template-processed image, obtaining a processed difference image between the template image and the template-processed image, wherein a difference image can be generated by subtracting the template-processed image from the template image, wherein a region with a zero pixel value indicates that the two images are the same, and a non-zero region indicates a processed region, and an image of the processed region in the template-processed image is selected as the processed difference image;

[0061] (4) graying the processed difference image, and using the image feature extraction module to process the grayed processed difference image to obtain corresponding difference features;

[0062] (5) setting a plurality of processing method labels corresponding to the image processing methods one by one, and associating the difference features with the corresponding processing method labels;

[0063] (6) Based on the difference features, the template-processed image and the image processing label, a first machine learning model is trained to obtain a processing method detection model. During training, the template-processed image is used as input, and the difference features and the image processing label are used as output, wherein the difference features are the output of the hidden layer, and the processing label is the output of the output layer. The processing method detection model is used to detect the difference features in the processed image based on the input processed image, obtain the pixel coordinates of the difference features in the processed image to obtain the difference coordinates, and output the processing method label and the difference coordinates associated with the difference features.

[0064] The present invention is not limited to specific machine learning models, such as convolutional neural networks (CNN), support vector machines (SVM), random forests (Random Forest) generative adversarial networks (GANs), etc.

[0065] The image processing method recommendation module is used to obtain the image features of the original image corresponding to the processed image at the difference coordinate position of the processed image, obtain the processing position feature of the original image, train the second machine learning model based on the original image, the image processing method and the processing position feature, and obtain the processing method recommendation model. The processing method recommendation model is used to obtain the recommended image processing method based on the input original image, detect the processing position feature, and output the processing position coordinates. The processing position coordinates are the pixel coordinates of the processing position feature. The pixel coordinates in the original image can be directly obtained through the detected processing position feature, including the following steps:

[0066] (1) Selecting an original image for which a corresponding processed image exists;

[0067] (2) obtaining image processing methods of the original image and the corresponding processed image, and obtaining a processing method label corresponding to the image processing method;

[0068] (3) Inputting the original image into the image feature extraction module to obtain the corresponding original image features;

[0069] (4) Based on the original image features, the processing method label and the processing position features, a second machine learning model is trained to obtain a processing method recommendation model. During training, the original image features are used as input, and the processing position features and the processing method label are used as output, wherein the processing position features are the hidden layer outputs, and the processing method label is the output layer output. The model is used to extract the original image features from the input original image through the image feature extraction module, retrieve the corresponding processing position features based on the original image features, obtain the pixel coordinates of the processing position features to obtain the processing position coordinates, and output the corresponding processing method label and processing position coordinates.

[0070] The system also includes a processing method verification module, which is used for the user to select a user to verify that the relationship is correct, and the processed image and the corresponding processing method label and pixel coordinates are saved as samples.

[0071] The processing method recommendation module is also used to use the samples to train and update the processing method detection model when the samples reach a set first sample quantity.

[0072] The image processing parameter recommendation module is used to analyze the difference between the processing position characteristics of each original image and the difference characteristics of the corresponding processed image, obtain the difference characteristic parameters, and train the third machine learning model based on the original image, the image processing method and the difference characteristic parameters to obtain the processing parameter prediction model. The processing parameter prediction model is used to output the predicted difference characteristic parameters based on the input original image and the image processing method, including the following steps:

[0073] (1) Input the original image into the image feature extraction module to obtain the corresponding original image features;

[0074] (2) classifying the original image and the corresponding processed image according to different processing methods to obtain multiple processing groups;

[0075] (3) extracting the processing position features and difference features corresponding to the processing methods corresponding to the processing groups in each processing group, respectively, and obtaining the difference feature parameters of the difference features and the corresponding difference position features;

[0076] (4) training a third machine learning model based on the original image features and difference feature parameters of each processing group to obtain a processing parameter prediction sub-model, wherein the original image features are used as input and the difference feature parameters are used as output during training;

[0077] (5) The processing parameter prediction sub-models of all processing groups are merged to obtain a processing parameter prediction model. The processing parameter prediction model selects the corresponding processing parameter prediction sub-model according to the input processing method label, extracts the original image features through the image feature extraction module, and outputs the corresponding difference feature parameters based on the original image features.

[0078] The image processing module is used to obtain a new original image, input the new original image into the processing method recommendation model, obtain the corresponding image processing method and processing position coordinates, and then input the new original image and the obtained image processing method into the processing parameter prediction model to obtain the corresponding difference feature parameters, and process the new original image based on the obtained image processing method, processing position coordinates and difference feature parameters to obtain the corresponding new processed image.

[0079] See also Figure 2 , an AI image processing method, comprising the following steps:

[0080] S1, select the original image and the corresponding processed image through the image collection module and save them for later use;

[0081] S2, extracting features from the original image and the processed image through an image feature extraction module;

[0082] S3, training the first machine learning model through the image processing method detection module to obtain a processing method detection model;

[0083] S4, training the second machine learning model through the image processing method recommendation module to obtain a processing method recommendation model;

[0084] S5. Train the third machine learning model through the image processing parameter recommendation module to obtain a processing parameter prediction model;

[0085] S6. The new original image to be processed is inputted through the image collection module, and the image processing module processes the new original image to obtain a new processed image through the processing method recommendation model and the processing parameter prediction model.

[0086] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An AI image processing system, characterized in that: It includes an image collection module, an image feature extraction module, an image processing method detection module, an image processing method recommendation module, an image processing parameter recommendation module, and an image processing module; The image collection module is used to acquire images, select an unprocessed image to save, obtain an original image, and select a processed image corresponding to the original image to save; The image feature extraction module is used to extract features from the original image and the processed image; The image processing method detection module is used to select an original image as a template image, and set a plurality of image processing methods, respectively use the image processing methods to process the template image, obtain a corresponding template processing image, analyze the difference between the template processing image and the corresponding template image, obtain the difference characteristics of each image processing method, train a first machine learning model based on the template processing image, the difference characteristics, and the image processing method, obtain a processing method detection model, the processing method detection model is used to identify the difference characteristics in the input processing image and the pixel coordinates of the difference characteristics in the processing image, obtain the difference coordinates, and output the corresponding image processing method and the difference coordinates; The image processing method recommendation module is used to obtain the image features of the original image corresponding to the processed image at the difference coordinate position of the processed image, obtain the processing position feature of the original image, train the second machine learning model based on the original image, the image processing method and the processing position feature, and obtain the processing method recommendation model. The processing method recommendation model is used to obtain the recommended image processing method based on the input original image, detect the processing position feature, and output the processing position coordinates. The processing position coordinates are the pixel coordinates of the processing position feature, and its pixel coordinates in the original image can be directly obtained through the detected processing position feature; The image processing parameter recommendation module is used to analyze the difference between the processing position feature of each original image and the difference feature of the corresponding processed image to obtain the difference feature parameter, and train the third machine learning model based on the original image, the image processing method and the difference feature parameter to obtain the processing parameter prediction model, and the processing parameter prediction model is used to output the predicted difference feature parameter based on the input original image and the image processing method; The image processing module is used to obtain a new original image, input the new original image into the processing method recommendation model to obtain the corresponding image processing method and processing position coordinates, and then input the new original image and the obtained image processing method into the processing parameter prediction model to obtain corresponding difference feature parameters, and process the new original image based on the obtained image processing method, processing position coordinates and difference feature parameters to obtain a corresponding new processed image.

2. The AI ​​image processing system according to claim 1, characterized in that: The image processing mode detection module is used to select an original image as a template image, and set a plurality of image processing modes, respectively use the image processing modes to process the template images, obtain corresponding template processing images, analyze the difference between the template processing images and the corresponding template images, obtain the difference features of each image processing mode, and train a first machine learning model based on the template processing images, the difference features, and the image processing modes to obtain a processing mode detection model, including the following steps: Select the original image as the template image and set various image processing methods; Using each image processing method to process the template image respectively to obtain a corresponding template processed image; Based on the template image and the template-processed image, a processed difference image of the template image and the template-processed image is obtained; graying the processed difference image, and using the image feature extraction module to process the processed difference image after the graying process to obtain corresponding difference features; Setting a plurality of processing method labels to correspond to image processing methods one by one, and associating the difference features with the corresponding processing method labels; Based on the difference features, template-processed images and image processing labels, a first machine learning model is trained to obtain a processing method detection model. During training, the template-processed image is used as input, and the difference features and image processing labels are used as output. The processing method detection model is used to detect the difference features in the processed image based on the input processed image, obtain the pixel coordinates of the difference features in the processed image to obtain the difference coordinates, and output the processing method label associated with the difference features and the difference coordinates.

3. The AI ​​image processing system according to claim 2, characterized in that: The system further comprises a processing mode verification module, wherein the processing mode verification module is used for the user to select the user to verify that the relationship is correct, and the processed image and the corresponding processing mode label and pixel coordinates are saved as samples.

4. The AI ​​image processing system according to claim 3, characterized in that: The processing method recommendation module is also used to use the samples to train and update the processing method detection model when the samples reach a set first sample quantity.

5. The AI ​​image processing system according to claim 1, characterized in that: The image processing method recommendation module trains a second machine learning model based on the original image, the image processing method and the processing position features to obtain a processing method recommendation model, wherein the processing method recommendation model is used to obtain a recommended image processing method based on the input original image, and comprises the following steps: Select an original image for which a corresponding processed image exists; Obtaining image processing methods of an original image and a corresponding processed image, and obtaining a processing method label corresponding to the image processing method; Input the original image into the image feature extraction module to obtain the corresponding original image features; Based on the original image features, the processing method labels and the processing position features, a second machine learning model is trained to obtain a processing method recommendation model. During training, the original image features are used as input, and the processing position features and the processing method labels are used as output. The model is used to extract the original image features from the input original image through an image feature extraction module, retrieve the corresponding processing position features based on the original image features, obtain the pixel coordinates of the processing position features to obtain the processing position coordinates, and output the corresponding processing method labels and the processing position coordinates.

6. The AI ​​image processing system according to claim 1, characterized in that: The image processing parameter recommendation module trains a third machine learning model based on the original image, the image processing method and the difference feature parameters to obtain a processing parameter prediction model, including the following steps: Input the original image into the image feature extraction module to obtain the corresponding original image features; Classifying the original image and the corresponding processed image according to different processing methods to obtain multiple processing groups; Extracting the processing position features and the difference features corresponding to the processing methods corresponding to the processing groups in each processing group, respectively, and obtaining difference feature parameters of the difference features and the corresponding difference position features; Based on the original image features and difference feature parameters of each processing group, a third machine learning model is trained to obtain a processing parameter prediction sub-model, wherein the original image features are used as input and the difference feature parameters are used as output during training; The processing parameter prediction sub-models of all processing groups are merged to obtain a processing parameter prediction model. The processing parameter prediction model selects the corresponding processing parameter prediction sub-model according to the input processing method label, extracts the original image features through the image feature extraction module, and outputs the corresponding difference feature parameters based on the original image features.

7. An AI image processing method, applicable to an AI image processing system according to any one of claims 1 to 6, characterized in that: The following steps are involved: S1, select the original image and the corresponding processed image through the image collection module and save them for later use; S2, extracting features from the original image and the processed image through an image feature extraction module; S3, training the first machine learning model through the image processing method detection module to obtain a processing method detection model; S4, training the second machine learning model through the image processing method recommendation module to obtain a processing method recommendation model; S5. Train the third machine learning model through the image processing parameter recommendation module to obtain a processing parameter prediction model; S6. A new original image to be processed is inputted through the image collection module, and the image processing module processes the new original image to obtain a new processed image through a processing method recommendation model and a processing parameter prediction model.

Images