Wound surface automatic pairing and color calibration method and system based on AI vision

Abstract

The invention relates to the field of artificial intelligence image processing, in particular to a method for realizing automatic matching and color calibration of color cards by utilizing an AI visual technology, which comprises the following steps of: firstly, positioning a homocentric-square-shaped mark of the color card based on a convolutional neural network and correcting the direction of a photo by combining a custom color card design and the AI visual technology; then extracting color features of a color card region, constructing a color mapping relation by adopting three-channel independent polynomial regression, and completing overall color correction; and finally, introducing an adaptive color correction neural network, and carrying out refined color calibration on the medical wound area. According to the method, the precision, the real-time performance and the robustness of size estimation and color calibration in the medical image are remarkably improved, and the method is suitable for scenes such as complex illumination and rotation and has important clinical application value.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence image processing, and more particularly to an AI vision-based wound automatic matching and color calibration method and system. Background Art

[0002] With the rapid development of computer vision and artificial intelligence technologies, image analysis, color correction, and true size estimation have shown widespread application value in medical imaging, trauma measurement, skin lesion recording, and other fields. In medical images or clinical photographs, accurately measuring the true size, color, and position of the target is crucial for clinical diagnosis, efficacy evaluation, and wound healing tracking. However, traditional methods usually rely on a combination of standard color cards and manual measuring rulers, and perform analysis through manual operation or simple image processing techniques. This method has significant shortcomings: manual measurement is inefficient, the process is cumbersome and time-consuming, and it is difficult to meet the needs of efficient batch processing. It is also easily affected by subjective factors, resulting in large errors. In addition, traditional image processing methods such as template matching and edge detection are sensitive to image rotation, tilt, scale changes, and changes in lighting conditions, and are prone to inaccurate positioning, insufficient precision, and poor robustness, which seriously affects the actual application effect.

[0003] In recent years, some studies have attempted to use algorithms based on feature point detection (such as SIFT, SURF) or template matching methods to achieve color card positioning and size calibration, but these methods still have obvious limitations. First, feature point detection algorithms perform poorly when dealing with situations with single texture, complex lighting or image rotation, and the detection accuracy drops significantly; second, traditional template matching methods are less robust to scale changes and rotation transformations, resulting in low matching accuracy; third, these methods have high computational complexity, and their real-time and efficiency performance cannot meet the high requirements of medical imaging scenarios for computing speed and accuracy.

[0004] Based on the public number CN120221096A A method for constructing a wound grading model and a wound self-analysis system, with a publication date of June 27, 2025. The patent specially designs the color card by combining it with the wound, and performs visual correction based on the photographed designed color card, and applies the corrected color card and wound image to wound analysis. However, the technical solution does not involve the technical means of how to correct the color card and the wound. In addition, the technical solution does not involve intelligent correction and the combination of medical wound-specific color correction, which may lead to color deviation and affect the accuracy of clinical judgment. Therefore, there is an urgent need for a technical solution that can overcome the above-mentioned defects to improve the accuracy and real-time performance of photo size estimation and color correction in actual scenes such as medical images. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the purpose of the present invention is to provide an AI vision-based automatic wound matching and color calibration method and system, aiming to solve the problems of poor positioning accuracy, high computational complexity, and insufficient real-time performance of traditional methods in complex scenarios such as rotation and tilt. By introducing AI algorithms and custom color card design, the present invention not only significantly improves the positioning success rate and color correction accuracy, but also enhances the system's adaptability to complex environments, providing more reliable technical support for medical image analysis and clinical diagnosis, and has important application value and technical advantages in improving efficiency, accuracy and robustness.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An AI vision-based automatic wound matching and color calibration method includes the following steps:

[0008] An image acquisition step is to obtain an image including the color card and the wound surface captured by a visual camera as an image to be processed;

[0009] an overall image correction step, determining a deflection angle of the color card according to the marking points in the color card, and deflecting the image to be processed based on the deflection angle to obtain a corrected image;

[0010] an image color difference processing step, extracting a color area from the corrected image, mapping the color of each color block in the color area with the color of each color block in a preset standard color card to obtain a color difference, and performing color correction on the corrected image based on the color difference to obtain an image to be analyzed;

[0011] The wound area color correction step extracts the wound area from the image to be analyzed, introduces an adaptive color correction neural network to construct a medical wound-specific color correction mapping function, and performs final calibration on the pixels of the wound area.

[0012] Furthermore, the wound area color correction step includes a network construction strategy, which includes a dual-channel feature input step, a feature extraction step, and a channel processing step;

[0013] The dual-channel feature input step constructs an ACCNet network with dual-channel input, wherein the first channel inputs the RGB features of 22 color blocks of the standard color card, and the second channel inputs the color distribution histogram features of the wound area;

[0014] In the feature extraction step, the dual-channel features are processed through a feature extraction convolution layer to obtain a feature map;

[0015] The channel processing step weights the feature map through a channel attention mechanism to highlight the wound-specific color channel information, and locates the color focus area of ​​the wound area through a spatial attention mechanism to strengthen the attention to the color of the wound tissue.

[0016] Furthermore, the wound area color correction step includes a mapping function generation strategy, and the mapping function generation strategy includes a correction parameter output step and a loss function setting step;

[0017] The correction parameter output step outputs a color correction parameter vector of the wound area through the ACCNet network, wherein the correction parameter vector includes a correction weight for each color channel;

[0018] In the loss function setting step, the loss function of the ACCNet network is set to:

[0019] ;

[0020] in, Color calibration error for color cards, For clinical realism loss, is the balance coefficient, Including typical values ​​of healthy tissue, typical values ​​of granulation tissue, typical values ​​of fibrinous tissue and typical values ​​of necrotic tissue.

[0021] Furthermore, the image color difference processing step includes a color mapping strategy, which includes constructing a color mapping relationship between each color block color and each color block color in a preset standard color card through a three-channel independent polynomial regression method to obtain color difference.

[0022] Furthermore, the three-channel independent polynomial regression method includes a color channel normalization step, a polynomial model construction step, and a color mapping function generation step;

[0023] The color channel normalization step is to normalize the RGB channel values ​​of the color card in the corrected image with the RGB channel values ​​of a preset standard color card;

[0024] The polynomial model construction step constructs an n-order polynomial mapping model for each channel;

[0025] The color mapping function generation step solves the polynomial parameters of each channel by the normal equation method with the goal of minimizing the loss function, and constructs a color mapping function according to the polynomial parameters of each channel.

[0026] Furthermore, the image color difference processing step also includes a color region extraction strategy, and the color region extraction strategy includes a color region cutting step and a grid center point positioning step;

[0027] In the color area cutting step, identify the coordinates of three marker points of the color card in the corrected image, calculate the coordinates of the four top corners of the color card area, and cut the color area of the color card through the edge contour algorithm according to the coordinates of the four top corners;

[0028] In the grid center point positioning step, divide the color area into grids, calculate the size of each grid, and determine the pixel position of the center point of each grid.

[0029] Furthermore, the color area extraction strategy further includes a marker point grid exclusion step and a color matching relationship establishment step;

[0030] In the marker point grid exclusion step, exclude the grids where the three "square inside a square" marker points in the color area fall;

[0031] In the color matching relationship establishment step, extract the RGB three-channel color values of the center point pixels of the remaining grids, obtain the standard color values of the corresponding grids in the preset standard color card, and establish a one-to-one matching relationship between the two.

[0032] Furthermore, in the overall image correction step, there is a correction strategy, and the correction strategy includes a marker point recognition step, a center point coordinate detection step, and a rotation transformation step;

[0033] In the marker point recognition step, recognize at least three marker points in the color card. The marker points are "square inside a square" patterns with black and white alternation, and are located at the upper left, upper right, and lower left positions of the color card area respectively;

[0034] In the center point coordinate detection step, detect the coordinate center points of the "square inside a square" patterns through a convolutional neural network to obtain three center point coordinates;

[0035] In the rotation transformation step, calculate the deflection angle of the color card based on the line segment formed by the upper left and upper right marker points, and transform the待处理图像 (to-be-processed image) through the rotation transformation formula based on the deflection angle to obtain the corrected image.

[0036] Furthermore, in the wound area color correction step, obtain the corrected color value of each pixel in the wound area of the待分析图像 (to-be-analyzed image) after image color difference processing, perform calibration on each pixel with the calibration parameter vector, and then restore the calibrated color value to the range of 0-255 to obtain the finally calibrated wound area image.

[0037] An AI vision-based wound automatic pairing and color calibration system includes:

[0038] An image acquisition module, which acquires an image including a color card and a wound taken by a vision camera as a to-be-processed image;

[0039] An overall image correction module determines a deflection angle of the color card according to the marking points in the color card, and deflects the image to be processed based on the deflection angle to obtain a corrected image;

[0040] An image color difference processing module extracts a color area from the corrected image, maps the color of each color block in the color area to the color of each color block in a preset standard color card to obtain a color difference, and performs color correction on the corrected image based on the color difference to obtain an image to be analyzed;

[0041] The wound area color correction module extracts the wound area from the image to be analyzed, introduces an adaptive color correction neural network to construct a medical wound-specific color correction mapping function, and finally calibrates the pixels of the wound area.

[0042] The beneficial effects of the present invention are as follows: 1. By combining the custom color card design with AI visual technology, a significant improvement in the size estimation and color calibration technology in medical images and clinical photos is achieved. In particular, the method preliminarily calibrates the overall color difference through traditional color card pixel mapping, and the AI ​​model then performs fine color calibration of specific areas of the medical wound surface, achieving multi-level precision correction from the overall to the local level. Among them, the typical color benchmarks of medical wound tissue (such as granulation, necrosis, and yellow tissue) are introduced as guidance for AI model training, so that the color calibration is more in line with clinical medical diagnosis needs. Furthermore, the method adaptively learns the characteristics of medical wound surfaces, effectively responds to complex factors such as lighting, angles, and tissue heterogeneity under different shooting conditions, and greatly enhances the robustness of the system. In particular, the AI ​​model can adaptively adjust the correction parameters in real time based on the medical color characteristics of the wound image itself, avoiding the defects of the traditional single mapping method with fixed parameters and difficult real-time adjustment. Combining traditional pixel correction with AI medical benchmark constraints, the color of the corrected image is closer to the real wound color expected by clinical experts, which is conducive to improving the accuracy of diagnosis and treatment decisions;

[0043] 2. Using polynomial regression combined with AI color correction in the RGB three-channel approach, the root mean square error (RMSE) between each color block of the photo color card and the corresponding block of the standard color card was reduced to an average of less than 2.6 pixel values, which is more than 70% lower than traditional manual or linear correction methods (the error is usually within the range of 10-15 pixel values). The color structure similarity (SSIM) between the corrected photo color card and the standard color card was increased to above 0.97, much higher than traditional methods (generally around 0.85), effectively improving the color authenticity of the photo. By introducing AI algorithms and custom color card positioning marks, the present invention has enhanced its adaptability to rotation, tilt, and complex lighting environments, achieving a positioning success rate of over 98%, significantly reducing the risk of color correction failure caused by positioning errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a step diagram of the calibration method of the present invention;

[0045] Figure 2 is a flow chart of color correction of an image to be processed in the present invention;

[0046] Figure 3 It is the color card structure diagram of the present invention; Figure 4 This is a sample diagram of the cutting contour obtained in the present invention. DETAILED DESCRIPTION

[0047] The present invention will be described in further detail below with reference to the accompanying drawings and embodiments. Identical components are denoted by the same reference numerals. It should be noted that the terms "front," "rear," "left," "right," "upper," and "lower" used in the following description refer to directions in the accompanying drawings, and the terms "bottom," "top," "inner," and "outer" refer to directions toward or away from the geometric center of a particular component, respectively.

[0048] This invention provides a method and system for automatic color chart matching and color calibration using AI vision technology. Through custom color chart design, AI vision technology for positioning and color feature extraction, and two-stage color correction using polynomial regression combined with an adaptive neural network, this method significantly improves the accuracy and real-time performance of size estimation and color calibration in medical images and clinical photographs. Specific embodiments of the invention are described in detail below with reference to specific figures in the accompanying description of the drawings.

[0049] like Figure 1 The figure shows the steps of the calibration method. The entire method is divided into several key steps, including custom color card design, image acquisition, color card positioning and correction, grid construction, color mapping relationship establishment, overall color correction and fine-grained calibration of the wound area. Each step has been fully verified in actual application scenarios and has a clear technical implementation path.

[0050] First, based on the applicant's application for publication number CN120221096AA method for constructing a wound grading model and a wound self-analysis system, a custom color card is designed. The color card consists of two parts: a color card area and a handheld area. The color card area adopts a 5×5 grid layout, with a total of 25 squares. Among them, 3 squares are black-and-white "hui" - shaped positioning marks, located at the upper left, upper right, and lower left corners respectively. The remaining 22 squares are standard color blocks for color feature extraction. The handheld area facilitates the user's handheld operation to avoid contamination or occlusion of the color card area. The uniqueness of this color card design lies in achieving precise positioning and angle correction of the color card in the image through the "hui" - shaped positioning marks, and at the same time providing a precise benchmark for subsequent color correction through 22 standard color blocks. In practical applications, it is required that the user takes a photo containing the target wound and the custom color card, and ensures that the photo resolution is , where represents the width (pixels), represents the height (pixels). When taking the photo, it is necessary to ensure that the color card and the wound are on the same plane and there is no obvious occlusion to improve the accuracy of subsequent processing.

[0051] Then enter the image acquisition and preliminary processing stage, obtain the image captured by the vision camera including the color card and the wound as the image to be processed. In order to achieve precise positioning and orientation correction of the color card, a convolutional neural network model is used to detect the positions of the three "hui" - shaped positioning marks in the color card area. As Figure 3 shown, it is an example of obtaining the "hui" - shaped coordinate points. The convolutional neural network model can effectively identify the center point coordinates of the "hui" - shaped pattern under complex lighting conditions through training on a large number of labeled data:

[0052] ,

[0053] According to three points, determine the rotation angle of the color card in the photo . Set the line segment formed by the "hui" - shaped patterns 1 and 2 as the reference line, then the angle calculation formula is: ; According to the calculated rotation angle , perform a rotation transformation on the image to be processed to obtain the corrected image ,

[0054] ,

[0055] The core of this step is to improve the robustness and accuracy of color card positioning through AI algorithms, especially maintaining a high positioning success rate under complex lighting conditions.

[0056] After completing the color card positioning, enter the stage of precise cutting and grid construction of the color card area; According to the rotation angle and the "hui" - shaped coordinates, calculate the four vertex coordinates of the color card , and determine the area of the color card according to the coordinate points as:

[0057] ,

[0058] At the same time, the color card area is divided into a 5×5 grid, and the size of each square is: , then the th row, the th column square center point pixel position:

[0059] ,

[0060] Such as Figure 4 shown for obtaining the cutting contour sample. This process is implemented through the edge contour algorithm to ensure the precise cutting of the color card area, calculate the size of each grid, and determine the center point pixel position of each grid.

[0061] Specifically, after excluding the three "hui"-shaped positioning marks in the upper left, upper right, and lower left in the image color card area , the RGB three-channel color values of the centers of the remaining 22 standard color squares are extracted. Among them, the center point pixel coordinates of the remaining 22 standard color squares are:

[0062] { },

[0063] ​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​

[0070] Get the standard color card color vector: ,

[0071] Then match the color vector of the image color card with the color vector of the standard color card one by one to establish the color mapping relationship and position matching relationship. Color square positions Block Location ;

[0072] Finally, we get a strict one-to-one color matching: ,During the color feature extraction process, the RGB values ​​are normalized and mapped to the [0,1] interval to eliminate the brightness differences that may exist between different devices.

[0073] Then, we enter the color mapping correction relationship construction stage. The present invention uses a three-channel independent polynomial regression method to establish the color mapping relationship and generate a color correction function. Each color channel value is normalized to the interval [0, 1]. The normalized RGB channel of the photo color card is represented as: ,

[0074] Standard color card RGB channel normalization representation: ,

[0075] The normalized color set is recorded as:

[0076] Photo color chart color: ,

[0077] Standard color card colors: ,

[0078] The color mapping model is defined as an n-order polynomial. For example, the mapping model of the R channel is:

[0079] ,

[0080] in =[ ] is the parameter set to be solved. Similarly, the G and B channel mapping functions are defined as:

[0081] ,

[0082] ,

[0083] The parameters of the polynomial regression model are solved by the least squares method, the optimization goal is clarified, and the loss functions are defined as follows: , specifically:

[0084] ,

[0085] ,

[0086] ,

[0087] Minimize the above loss functions respectively, and explicitly solve the parameters by the normal equation method. Taking the R channel as an example, define the input vector and parameter vector as:

[0088] , ,

[0089] The parameter solution is: , similarly, we can find the parameters and ;

[0090] Finally, we get the color mapping function of three channels ,

[0091] ,

[0092] For subsequent color correction, the color value of any pixel in the photo is normalized, a color correction function is applied, and the corrected color value is restored to the range of 0-255. This step ensures the consistency and authenticity of the overall color of the image by calibrating the RGB value of each pixel one by one.

[0093] Specifically, for any pixel in the photo The original color value ( ), perform the following steps to calibrate:

[0094] First, normalize the original color values: ,

[0095] Then, apply the color correction function: ,

[0096] Next, restore the corrected color value to the range of 0~255:

[0097] ,

[0098] Finally, perform the above steps on all pixels in the photo to obtain the corrected image with accurate color correction. .

[0099] After completing the overall color correction, the adaptive color correction neural network (ACCNet) was further introduced to perform fine calibration on the wound surface images that had completed color correction based on the color card. The network uses a dual-channel feature input structure: the first channel inputs the standard color card features (color reference), including the RGB features of the 22 color blocks of the color card; the second channel inputs the wound surface features, using the color histogram features of the wound surface image, color distribution histogram: ,in Indicates the wound picture The RGB color value of each pixel is obtained, where N is the number of effective pixels in the image; the two channel features reflect the medical tissue-specific characteristics of wound color, and are fused through the feature extraction convolution layer and the attention module respectively: the channel attention module is used to highlight the wound-specific color channel information and automatically increase the attention weight of wound features; the spatial attention module is used to accurately locate the wound color space area that the color correction model should focus on; ACCNet outputs an adaptive color correction parameter vector, including the correction weight of each color channel (RGB), and introduces the color authenticity constraint of medical wound images through the loss function design.

[0100] The color correction of the wound area includes a mapping function generation strategy. Specifically, the ACCNet network outputs an adaptive color correction parameter vector , including correction weights for each color channel (RGB): ;

[0101] The loss function design of ACCNet introduces color authenticity constraints for medical wound images:

[0102] ,

[0103] in is the color calibration error of the color card (the error between the wound surface image and the standard color card), It is a color correction method for clinical wounds with loss of realism. The network correction is guided by pre-labeled real color benchmarks of wounds confirmed by clinical medical experts (such as typical medical colors such as red tissue, yellow tissue and black necrotic tissue).

[0104] In the present invention, , considering that wound color correction faces a variety of wounds, we set the following respectively:

[0105] Typical values ​​for healthy tissue =(210,165,155),

[0106] Typical values ​​of granulation tissue =(170,55,55),

[0107] Typical values ​​for cellulose tissue =(195,160,55),

[0108] Typical values ​​of necrotic tissue =(45,35,30),

[0109] is the balance coefficient, which is 0.48 in the present invention.

[0110] Implementation of precise color correction of wound images: AI-based precise calibration of the original RGB color value (R, G, B) of each pixel of the medical wound image.

[0111] ,

[0112] Finally, the final calibrated wound area image is obtained .

[0113] In practical applications, this method can be embedded in a variety of hardware devices for end-side deployment, such as portable medical image analyzers and smart terminals. Taking embedded devices as an example, the device side needs to be equipped with a high-performance processor (such as GPU or TPU) to support the fast reasoning of convolutional neural networks. At the same time, it needs to have sufficient storage space to save pre-trained models and intermediate data. The recommended training set size is more than 2,000 high-quality medical images, covering a variety of shooting conditions (such as different lighting, angles, resolutions, etc.). In terms of hyperparameter setting, the preferred range of learning rate is 0.001-0.01, and the preferred range of batch size is 32-128. The number of network layers can be adjusted according to specific task requirements. During actual operation, the system first loads the pre-trained model and initializes the parameters, and then receives the images to be processed uploaded by the user. By executing the above steps, the system can complete the entire process from image acquisition to final color correction in milliseconds, meeting real-time requirements.

[0114] Based on the wound color calibration method, a calibration system was designed, including:

[0115] An image acquisition module acquires an image including the color card and the wound surface captured by a visual camera as an image to be processed;

[0116] An overall image correction module determines a deflection angle of the color card according to the marking points in the color card, and deflects the image to be processed based on the deflection angle to obtain a corrected image;

[0117] An image color difference processing module extracts a color area from the corrected image, maps the color of each color block in the color area to the color of each color block in a preset standard color card to obtain a color difference, and performs color correction on the corrected image based on the color difference to obtain an image to be analyzed;

[0118] The wound area color correction module extracts the wound area from the image to be analyzed, introduces an adaptive color correction neural network to construct a medical wound-specific color correction mapping function to perform final calibration on the pixels of the wound area.

[0119] This invention uses traditional color card pixel mapping to preliminarily calibrate the overall color difference, and the AI ​​model then performs fine color calibration of specific areas of the medical wound surface, achieving multi-level precise correction from the overall to the local level. It introduces typical color benchmarks of medical wound tissue (such as granulation, necrosis, and yellow tissue) as a training guide for the AI ​​model, making the color calibration more in line with clinical medical diagnosis needs. By adaptively learning the characteristics of medical wound surfaces, it effectively responds to complex factors such as lighting, angle, and tissue heterogeneity under different shooting conditions, thereby greatly enhancing the robustness of the system.

[0120] The AI ​​model adaptively adjusts correction parameters based on the medical color characteristics of the wound image itself, avoiding the limitations of traditional single-mapping methods, which often have fixed parameters and are difficult to adjust in real time. By combining traditional pixel correction with AI medical benchmark constraints, the corrected image color more closely resembles the actual wound color expected by clinical experts, improving the accuracy of diagnostic and treatment decisions.

[0121] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that improvements and modifications that do not depart from the principles of the present invention are within the scope of protection of the present invention.

Claims

1. An AI-based vision-based automatic wound matching and color calibration method, characterized by: The steps include: An image acquisition step is to obtain an image including the color card and the wound surface captured by a visual camera as an image to be processed; an overall image correction step, determining a deflection angle of the color card according to the marking points in the color card, and deflecting the image to be processed based on the deflection angle to obtain a corrected image; an image color difference processing step, extracting a color area from the corrected image, mapping the color of each color block in the color area with the color of each color block in a preset standard color card to obtain a color difference, and performing color correction on the corrected image based on the color difference to obtain an image to be analyzed; The wound area color correction step extracts the wound area from the image to be analyzed, introduces an adaptive color correction neural network to construct a medical wound-specific color correction mapping function, and performs final calibration on the pixels of the wound area.

2. The method for automatic wound matching and color calibration based on AI vision according to claim 1, characterized in that: The wound area color correction step includes a network construction strategy, which includes a dual-channel feature input step, a feature extraction step, and a channel processing step; The dual-channel feature input step constructs an ACCNet network with dual-channel input, wherein the first channel inputs the RGB features of 22 color blocks of the standard color card, and the second channel inputs the color distribution histogram features of the wound area; In the feature extraction step, the dual-channel features are processed through a feature extraction convolution layer to obtain a feature map; The channel processing step weights the feature map through a channel attention mechanism to highlight the wound-specific color channel information, and locates the color focus area of ​​the wound area through a spatial attention mechanism to strengthen the attention to the color of the wound tissue.

3. The method for automatic wound matching and color calibration based on AI vision according to claim 2, characterized in that: The wound area color correction step includes a mapping function generation strategy, which includes a correction parameter output step and a loss function setting step; The correction parameter output step outputs a color correction parameter vector of the wound area through the ACCNet network, wherein the correction parameter vector includes a correction weight for each color channel; In the loss function setting step, the loss function of the ACCNet network is set to: ； in, Color calibration error for color cards, Loss of clinical realism, is the balance coefficient, Including typical values ​​of healthy tissue, typical values ​​of granulation tissue, typical values ​​of fibrinous tissue and typical values ​​of necrotic tissue.

4. The method for automatic wound matching and color calibration based on AI vision according to claim 1 or 3, characterized in that: The image color difference processing step includes a color mapping strategy, which includes constructing a color mapping relationship between each color block color and each color block color in a preset standard color card through a three-channel independent polynomial regression method to obtain color difference.

5. The method for automatic wound matching and color calibration based on AI vision according to claim 4, characterized in that: The three-channel independent polynomial regression method includes a color channel normalization step, a polynomial model construction step, and a color mapping function generation step; The color channel normalization step is to normalize the RGB channel values ​​of the color card in the corrected image with the RGB channel values ​​of a preset standard color card; The polynomial model construction step constructs an n-order polynomial mapping model for each channel; The color mapping function generation step solves the polynomial parameters of each channel by the normal equation method with the goal of minimizing the loss function, and constructs a color mapping function according to the polynomial parameters of each channel.

6. The method for automatic wound matching and color calibration based on AI vision according to claim 5, characterized in that: The image color difference processing step also includes a color region extraction strategy, which includes a color region cutting step and a grid center point positioning step; In the color region cutting step, three marker point coordinates of the color card are identified in the corrected image, the four vertex coordinates of the color card region are calculated, and the color region of the color card is obtained by cutting through the edge contour algorithm according to the four vertex coordinates; In the grid center point positioning step, the color region is divided into grids, the size of each grid is calculated, and the pixel position of the center point of each grid is determined.

7. The method for automatic wound matching and color calibration based on AI vision according to claim 6, characterized in that: The color region extraction strategy further includes a marker point grid exclusion step and a color matching relationship establishment step; In the marker point grid exclusion step, the grids into which the three "回" - shaped marker points in the color region fall are excluded; In the color matching relationship establishment step, the RGB three - channel color values of the center point pixels of the remaining grids are extracted, the standard color values of the corresponding grids in the preset standard color card are obtained, and a one - to - one matching relationship between the two is established.

8. The method for automatic wound matching and color calibration based on AI vision according to claim 7, characterized in that: In the overall image correction step, there is a correction strategy, and the correction strategy includes a marker point recognition step, a center point coordinate detection step, and a rotation transformation step; In the marker point recognition step, at least three marker points in the color card are recognized. The marker points are black - and - white "回" - shaped patterns, which are located at the upper - left, upper - right, and lower - left positions of the color card region respectively; In the center point coordinate detection step, the coordinate center points of the "回" - shaped patterns are detected through a convolutional neural network to obtain three center point coordinates; In the rotation transformation step, the deflection angle of the color card is calculated based on the line segment formed by the upper - left and upper - right marker points, and the to - be - processed image is transformed through the rotation transformation formula based on the deflection angle to obtain a corrected image.

9. The method for automatic wound matching and color calibration based on AI vision according to claim 8, characterized in that: In the wound area color correction step, the corrected color value of each pixel in the wound area of the to - be - analyzed image after image color difference processing is obtained, each pixel is calibrated with the calibration parameter vector, and then the calibrated color value is restored to the range of 0 - 255 to obtain the finally calibrated wound area image.

10. An AI vision-based automatic wound matching and color calibration system, characterized by: Including: An image acquisition module that obtains an image including a color card and a wound taken by a vision camera as a to - be - processed image; An overall image correction module that determines the deflection angle of the color card based on the marker points in the color card, and deflects the to - be - processed image based on the deflection angle to obtain a corrected image; An image color difference processing module that extracts the color region in the corrected image, maps the color of each color block in the color region to the color of each color block in the preset standard color card to obtain a color difference, and performs color correction on the corrected image according to the color difference to obtain a to - be - analyzed image; A wound area color correction module that extracts the wound area in the to - be - analyzed image, and introduces an adaptive color correction neural network to construct a medical wound - specific color correction mapping function to finally calibrate the pixels in the wound area.

Images