Infrared thermal imaging image processing method for pain area

Abstract

The invention provides an infrared thermal imaging image processing method for a pain area, and relates to the field of image enhancement, and the method combines multi-scale adaptive contrast enhancement and a bidirectional circulation mechanism, captures global and local temperature features of an infrared image, constructs an iterative feature enhancement module through an incremental fusion strategy, and carries out the image enhancement through the iterative feature enhancement module. Different level features are fused in stages, learnable parameters are introduced to dynamically adjust the new and old feature fusion proportion, the adaptability and flexibility of the model are improved, and a staged training strategy is adopted, so that the model can learn low-level thermal image information and also can extract high-level structures and thermal anomaly semantic features. According to the method, the adaptability of the model to different types of focus areas is enhanced, and the actual demand of efficient and automatic processing of the infrared image is met.

Description

Technical Field

[0001] The present invention belongs to the technical field of image enhancement, and particularly relates to a method for processing infrared thermal imaging images of pain regions. Background Art

[0002] The processing of infrared thermal imaging images of pain regions has important value in medical applications such as chronic disease management, postoperative monitoring, and nerve function assessment. However, limited by factors such as the resolution of infrared thermal imaging devices, the complexity of skin surface temperature distribution, and environmental interference, thermal images often have problems such as low contrast, blurred edges, and many thermal noise points, which affect the accurate identification and diagnosis of pain regions. Existing image enhancement methods mostly focus on global brightness contrast adjustment, ignoring the enhancement requirements for detailed features such as temperature distribution patterns and regional thermal anomalies, and usually require manual intervention, making it difficult to adapt to automated clinical processes.

[0003] In recent years, deep learning has made remarkable progress in medical image enhancement. However, existing methods are mostly general designs, lacking a structural modeling and dynamic enhancement mechanism for "regional abnormal thermal features" in infrared thermal images, and it is difficult to balance the feature expression of global and local thermal distributions during the enhancement process, easily causing blurred boundaries or detail distortion in hot spot regions.

[0004] To address these problems, the present invention proposes a method for processing infrared thermal imaging images of pain regions, aiming to improve the automation and accuracy of image processing and solve the deficiencies of existing methods. Summary of the Invention

[0005] The present invention proposes a method for processing infrared thermal imaging images of pain regions. Through structural vector modeling, region-guided enhancement, multi-dimensional evaluation, and weight coupling mechanism, the self-adaptability, detail retention ability, and regional pertinence of image enhancement are improved. This method is applicable to the visual enhancement of temperature abnormal regions in thermal images, especially suitable for high-precision identification and enhancement of pain-related temperature abnormal regions in a medical environment.

[0006] The present invention aims to propose an infrared thermal imaging image enhancement model and provide a method for processing infrared thermal imaging images of pain regions, including the following steps.

[0007] S1. Construct an initial structure description, collect infrared thermal imaging images of the pain region, construct a preliminary abnormal region distribution map based on edge segmentation, and make a structural vector map.

[0008] S2. According to the structural vector map, extract information on the number, location, size, and edge blur degree of the abnormal region to form a structure-guided map.

[0009] S3. Cut the infrared thermal imaging image into multiple image blocks, and each image block is enhanced according to its local features and the structure-guided map.

[0010] S4. Calculate three indicators of structural similarity, edge continuity, and local contrast for the enhanced image block, calculate the quality score, and adjust the enhancement parameters of the next image block according to the score.

[0011] S5. Use the structural frequency curvature coupling to map the structure of the thermal imaging image, calculate the local frequency response, regional curvature value, and information entropy of the image, and construct a non-linear coupling mapping function.

[0012] S6. Merge each image block to obtain the overall enhanced thermal imaging image output.

[0013] Preferably, in step S1, construct an initial structure description, collect infrared thermal imaging images of normal and diseased pain regions, where the diseased regions include inflammatory pain, neuropathic pain, and postoperative traumatic pain, annotate the region of interest for each thermal imaging image, and the region of interest includes an abnormal temperature rise edge, a core hot spot area, and a peripheral transition zone. Based on edge segmentation, construct a preliminary abnormal heat distribution map and make a structure vector map of the pain region.

[0014] Preferably, in step S2, the construction method of the structure guidance map is as follows: Extract the spatial position information, area scale information, and boundary temperature gradient blur index of each pain region from the abnormal heat distribution map, and encode the multi-dimensional thermal features into a feature combination for structural modeling. Among them, the description unit of each pain region contains coordinate information indicating its position relationship in the infrared thermal imaging image, a scale parameter reflecting its abnormal heat diffusion range, and a blur parameter for measuring the clarity of the edge temperature transition. The set of structure vectors is used to construct the overall structure guidance map , and the specific formula is: ; In the formula, is the structure guidance map, represents the central coordinates of the th pain region in sequence, the radius and the edge blur degree , is the number of regions.

[0015] Preferably, in step S2, information such as the central coordinates, area scale, and edge temperature blur degree of each pain region in the image is vectorized and encoded, and organized into a graph structure expression form that can be used for region localization and structure guidance, namely, a structure guidance graph. This graph is used to provide spatial constraints for thermal anomaly features during the enhancement process, enabling each image patch to be associated with the corresponding abnormal region, thereby achieving the regional adaptability of the enhancement operation. Through the construction of the structure guidance graph, not only the key temperature distribution features in the image are retained, but also the spatial selectivity and direction accuracy of the enhancement process are improved, especially suitable for image enhancement tasks with regions having various thermal distribution patterns.

[0016] Preferably, in step S3, the method for cutting the colony image patches is as follows: S31. Based on a preset two-dimensional grid division strategy, the input infrared thermal imaging image is evenly divided in the spatial dimension to form a number of non-overlapping sub-image regions. The size of each sub-image region is the same in the row and column directions, and the division result covers the entire infrared image. Each image patch is set with a clear position index identifier for region-level matching and correspondence with the structure guidance graph, so as to cut the infrared image into multiple equal-sized image patches according to the preset grid size , and the size of each image patch is , where and respectively represent row and column indices; S32. The enhancement processing operation adopts an image patch enhancement strategy guided by feature response. This strategy dynamically adjusts the contrast enhancement amplitude and the retention intensity of thermal texture details of the image patch by analyzing the temperature distribution features, structure guidance information, and their spatial position relationship in the overall infrared thermal imaging image. The enhancement parameters are jointly determined based on the image statistical characteristics of the current patch and the features of the corresponding abnormal thermal region in the structure guidance graph. By processing each patch one by one, differential enhancement operations are applied to each infrared image patch respectively. The calculation formula of the enhancement function is: ; In the formula, is the th image patch of the original infrared thermal imaging image, is the structure guidance graph, is the enhancement parameter, is the enhancement function.

[0017] Preferably, in step S3, this step first divides the entire image into equal-sized sub-blocks based on a two-dimensional regular grid strategy to generate several image sub-blocks with spatial index attributes, ensuring that each image unit can be independently located and enhanced during subsequent processing. Subsequently, combined with the regional feature information provided in the structure guidance map, customized enhancement operations are performed on each image block. During the enhancement process, the brightness, contrast, or edge intensity processing parameters of the image block are dynamically adjusted using the structure matching relationship. This processing strategy can allocate different enhancement amplitudes in different image regions to achieve the effect of highlighting key details in the detailed regions and moderately suppressing non-target regions, thereby improving the structural clarity and information density of the overall image.

[0018] Preferably, in step S4, the quality score calculation method is as follows: S41. Calculate the structural similarity SSIM of each enhanced image block. By statistically extracting the grayscale mean and , grayscale variance, and the covariance between the two in the image blocks before and after enhancement, a comprehensive evaluation model based on brightness consistency, contrast preservation, and structural reconstruction degree is constructed. After normalization and suppression term adjustment, an evaluation value representing the structural consistency of the image block is generated, and its calculation formula is: ; In the formula, and are the averages of the image block and its enhanced image block, is the covariance of the image block and the enhanced image block, and are constants; S42. The edge continuity index is constructed by analyzing the difference degree of the edge gradients before and after the enhancement of the image block. First, extract the gradient magnitude map in each image block, record the gradient response values of each pixel position in the original image and the enhanced image respectively, and then perform point-by-point difference calculation on the response values before and after at the same position and average the entire image region to obtain the edge perturbation statistical value. Calculate the edge continuity after enhancement, and the calculation formula is: ; In the formula, is the edge continuity score of the image block, is the gradient of the th pixel of the image block , is the gradient of the th pixel of the enhanced image block , is the number of pixels calculated within this image block; S43. Calculate the luminance deviation value of each pixel in the image block from its neighborhood, and then normalize it to a standard contrast response, which is jointly determined by the pixel value, the gray mean value, and the standard deviation of the image block where the pixel is located. After performing the above processing on the image blocks before and after enhancement respectively, calculate the difference amplitude between the two, and perform statistical aggregation within the entire block area to calculate the local contrast improvement score. The calculation formula is: ; In the formula, is the local contrast improvement score of the image block, is for the image block the th pixel's gray value, is for the enhanced image block the th pixel's gray value, is the mean value of all pixels in the image block , is the mean value of all pixels in the enhanced image block ; S44. The fusion score takes the three indicators of structural similarity, edge continuity, and local contrast as inputs, and generates an overall image block quality evaluation value according to the multi-factor weighted mechanism, and feeds it back to the parameter adjustment module. Each sub-indicator is assigned a fixed weight coefficient according to its importance in the task. In the fusion process, a linear superposition strategy is adopted, and the scale unity between the indicators is also considered. The calculation formula is: ; In the formula, is the weight coefficient.

[0019] Preferably, in step S4, for each enhanced image block, three independent quality evaluation indicators of structural similarity, edge continuity, and local contrast are introduced, which are used to evaluate whether the enhanced image maintains structural consistency, whether the edge shape is damaged, and whether the contrast of the detail area is improved respectively. By calculating the statistical attributes of the image block, a quantitative scoring mechanism is established, and the three sub-scores are fused according to the task weights to form a total quality score. The fusion score can not only be used to evaluate the current enhancement effect, but also be used as a feedback signal to dynamically adjust the enhancement parameters of the subsequent image blocks, realizing the adaptive adjustment and stable optimization of the enhancement strategy, and improving the enhancement consistency and local detail quality.

[0020] Preferably, in step S5, the calculation method of the infrared thermal imaging image structure frequency curvature coupling mapping is: S51. Perform wavelet transform on the original infrared thermal imaging image, extract its sub-band responses in the horizontal, vertical, and diagonal directions, which respectively reflect the texture intensity changes of the infrared thermal imaging image in different directions. For each pixel position, statistically calculate the response amplitudes in the three-direction sub-bands, obtain its overall frequency response intensity through an amplitude combination strategy, and calculate the local frequency response of the image. , and the calculation formula for the high-frequency component of each pixel point through wavelet transform is: ; In the formula, are the wavelet coefficients of the infrared thermal imaging image in the horizontal, vertical, and diagonal directions respectively; S52. By obtaining the gradient values of the infrared thermal imaging image in the horizontal and vertical directions, further calculate its second-order derivatives, including the horizontal change rate, vertical change rate, and mixed change rate, and introduce a small regularization factor during the processing to calculate the local curvature value. , and the calculation formula based on the second-order derivative is: ; In the formula, is the local curvature value at the th pixel, is the gradient of the infrared thermal imaging image in the x direction, is the gradient of the infrared thermal imaging image in the y direction, is the second-order derivative of the infrared thermal imaging image in the x direction, is the mixed second-order derivative of the infrared thermal imaging image in the xy direction, is the second-order derivative of the infrared thermal imaging image in the y direction, is a small constant; S53. Establish a sliding window centered on each pixel, statistically calculate the gray value distribution of all pixels within the window, and calculate its probability distribution. Based on the probability density, construct an entropy value function, and add a very small bias during the construction of the probability density to calculate the information entropy. , and the calculation formula based on the local gray distribution is: ; In the formula, is the probability distribution of the gray value , is a constant; S54. The generation of the enhancement weight is based on three characteristic parameters: structural frequency, local curvature, and information entropy, and is comprehensively fused through a non-linear coupling mechanism. First, the above three types of features are subjected to scale normalization processing. Subsequently, control factors are allocated according to task requirements to adjust the influence degree of each parameter on the enhancement weight. Then, control factors are allocated according to task requirements to adjust the influence degree of each parameter on the enhancement weight. The enhancement parameter is generated through a non-linear coupling mapping, and the calculation formula is: ; In the formula, is the final enhancement weight.

[0021] Preferably, in step S5, image pixel-level features are collected from three complementary dimensions, including the frequency response reflecting the high-frequency detail activity, the curvature change characterizing the edge bending or corner characteristics, and the information entropy measuring the regional complexity. Each feature calculates the statistical response of the image through a preset window, and after scale unification and feature normalization processing, it is input into the non-linear mapping function to generate the pixel-level enhancement response weight. This weight is used to control the processing intensity of each pixel during the enhancement process, strengthen the response in the texture-rich area, suppress interference in the background smooth area, and thus improve the overall image structure recognition rate, detail restoration ability, and visual balance, meeting the requirements of high-precision image enhancement tasks.

[0022] Compared with the prior art, the present invention has the following technical effects: A processing method for infrared thermal imaging images of pain regions provided by the present invention realizes the quantitative expression of the position, scale, and edge blurriness of abnormal heat regions in the image by constructing a pain region structure vector map, providing accurate regional guidance for subsequent image block enhancement. Different from the traditional method based on a unified enhancement strategy, the present invention supports local enhancement according to the differences of image blocks, and the enhancement result is more delicate and the boundary retention is more complete. A multi-dimensional quality scoring mechanism is introduced to dynamically evaluate each image block comprehensively based on structural similarity, edge continuity, and local contrast, and the result is fed back for subsequent enhancement parameter adjustment to construct a closed-loop adaptive enhancement process. Further, frequency response, curvature information, and information entropy are fused to construct a coupling mapping to generate pixel-level enhancement weights, realizing the dynamic balance between detail region enhancement and background region suppression, and improving the structural clarity and visual consistency of the enhanced image. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flowchart of infrared thermal imaging image data processing provided by the present invention.

[0024] Figure 2 is a diagram of the cutting and enhancement module of the infrared thermal imaging image block provided by the present invention.

[0025] Figure 3It is the flowchart of the infrared thermal imaging image block enhancement parameters provided by the present invention.

[0026] Figure 4 It is the module diagram of the infrared thermal imaging image non - linear coupling mapping function provided by the present invention.

[0027] Figure 5 It is the effect diagram of the infrared thermal imaging image before enhancement provided by the present invention.

[0028] Figure 6 It is the effect diagram of the enhanced image obtained by processing through the infrared thermal imaging image enhancement model provided by the present invention. Specific embodiments

[0029] The present invention aims to propose a processing method for infrared thermal imaging images of pain regions, propose an infrared thermal imaging image enhancement model, construct an infrared thermal image dataset, collect infrared images of normal and diseased pain regions, and perform region - of - interest annotation. The annotation content includes the center, area scale, and edge temperature information of the abnormal heat region. The annotated dataset is made for model training. The local contrast of the image is enhanced through a multi - scale mechanism. The image registration operation is based on transformation - invariant features to ensure the alignment of key thermal abnormal regions between images, reduce geometric differences, and improve the stability of the model under different shooting angles or thermal sensitivity resolution conditions. Combining local frequency response, curvature value, and information entropy features, the enhancement parameters are dynamically adjusted through structural frequency - curvature coupling mapping. The local frequency response captures high - frequency texture details in the thermal image through wavelet transform, the curvature value detects edge structure changes through the second - derivative, and the information entropy is used to evaluate the temperature complexity of the local region. Finally, the enhancement weight is generated through non - linear coupling mapping to achieve fine control of image details, retain structural features, and significantly improve the detail expression ability and visual clarity of infrared images.

[0030] Please refer to Figure 1 as shown in, a processing method for infrared thermal imaging images of pain regions in the embodiments of the present application.

[0031] S1. Construct an initial structure description, collect infrared thermal imaging images of the pain region, construct a preliminary abnormal region distribution map based on edge segmentation, and make a structure vector map.

[0032] Furthermore, in step S1, collect infrared thermal imaging images of normal and diseased pain regions. The diseased images include inflammatory pain, neuropathic pain, and postoperative traumatic pain regions. Use a professional medical image annotation tool to perform region - of - interest annotation on each thermal imaging image. The annotation content includes abnormal temperature edges, core hot spots, and transition temperature regions to ensure the precise positioning of key regions. The annotation results are saved in the XML standard format and are in one - to - one correspondence with the original image files to ensure the high quality and integrity of the infrared image dataset.

[0033] S2. Extract the information on the number, location, size, and edge blurriness of the abnormal regions from the structural vector diagram to form a structure guidance diagram.

[0034] Further, in step S2, the specific steps of the method for constructing the structure guidance diagram are as follows.

[0035] Extract the spatial location information, area scale information, and boundary temperature gradient blurriness index of each pain region from the abnormal heat distribution diagram, and encode the multi-dimensional heat features into a feature combination for structural modeling. Among them, the description unit of each pain region contains coordinate information indicating its position relationship in the infrared thermal imaging image, a scale parameter reflecting its abnormal heat diffusion range, and a blurriness parameter for measuring the clarity of the edge temperature transition. The structural vector set is used to construct the overall structure guidance diagram , and the specific formula is: ; In the formula, is the structure guidance diagram, successively represents the central coordinates of the th pain region, the radius , and the edge blurriness , is the number of regions. During the implementation process, the number of regions is controlled not to be greater than 50.

[0036] S3. Cut the infrared thermal imaging image into multiple image blocks, and each image block is enhanced according to its local features and the structure guidance diagram.

[0037] Further, in step S3, as Figure 2 shown, the specific steps for cutting the colony image blocks are as follows.

[0038] S31. Based on a preset two-dimensional grid division strategy, uniformly divide the input infrared thermal imaging image in the spatial dimension to form a number of non-overlapping sub-image regions. The size of each sub-image region is the same in the row and column directions, and the division result covers the entire infrared image. Each image block is set with a clear position index identifier for region-level matching and correspondence with the structure guidance diagram, so as to cut the infrared image into multiple equal image blocks according to the preset grid size , and the size of each colony image block is , where and respectively represent the row and column indices. During the implementation process, the length and width of the grid are controlled not to be greater than 100, and the initial value is 50×50; S32. The enhancement processing operation adopts an image block enhancement strategy guided by feature response. This strategy dynamically adjusts the contrast enhancement amplitude and the intensity of retaining thermal texture details of the image block by analyzing the temperature distribution characteristics, structural guidance information of the current infrared image block, and their spatial position relationship in the overall infrared thermal imaging image. The enhancement parameters are jointly determined based on the image statistical characteristics of the current block and the characteristics of the corresponding abnormal thermal region in the structural guidance map. Through the method of processing block by block, differential enhancement operations are applied to each infrared image block respectively. The calculation formula of the enhancement function is: ; In the formula, is the th image block of the original infrared thermal imaging image, is the structural guidance map, is the enhancement parameter with a value range of [0.5, 2.0], is the enhancement function.

[0039] S4. Calculate three indicators of structural similarity, edge continuity, and local contrast for the enhanced image blocks, calculate the quality score, and adjust the enhancement parameters of the next image block according to the score.

[0040] Further, in step S4, as Figure 3 shown, the calculation method of the quality score is as follows.

[0041] S41. Calculate the structural similarity SSIM of each enhanced image block. By statistically extracting the gray mean and , gray variance, and the covariance between the two before and after enhancement, construct a comprehensive evaluation model based on brightness consistency, contrast retention, and structural reconstruction degree. After normalization and suppression term adjustment, generate an evaluation value representing the structural consistency of the image block. Its calculation formula is: ; In the formula, and are the averages of the image block and its enhanced image block, is the covariance of the image block and the enhanced image block, that is , and are constants mainly used to avoid division by zero errors during the calculation process. In this implementation process , ; S42. The edge continuity index is constructed by analyzing the difference degree of edge gradients before and after image block enhancement. First, the gradient magnitude map is extracted in each image block, and the gradient response values at each pixel position in the original image and the enhanced image are respectively recorded. Subsequently, the point-by-point difference calculation is performed on the response values before and after at the same position, and the average value is obtained for the entire infrared thermal imaging image area to obtain the edge perturbation statistic value. The edge continuity after enhancement is calculated, and the calculation formula is: ; In the formula, is the edge continuity score of the image block, which is used to measure the coherence of the edge after image block enhancement, is the th pixel of the image block, representing the edge intensity of the pixel. When calculating the gradient, the Sobel operator is used to calculate the gradients in the horizontal and vertical directions, is the gradient of the th pixel of the enhanced image block, is the number of pixels calculated within the image block. In this implementation process, N ≤ 10000; S43. Calculate the brightness deviation value of each pixel in the image block from its neighborhood, and then normalize it to the standard contrast response. This response is jointly determined by the pixel value, the gray mean value, and the standard deviation of its image block. After performing the above processing on the image blocks before and after enhancement respectively, calculate the difference amplitude between the two, and perform statistical aggregation within the entire area to calculate the local contrast enhancement score. The calculation formula is: ; In the formula, is the local contrast enhancement score of the image block, which is used to measure the degree of contrast enhancement of the enhanced image in the local area, is the th gray value of the pixel of the image block, is the th gray value of the pixel of the enhanced image block, is the mean value of all pixels in the image block , , is the mean value of all pixels in the enhanced image block ; S44. The fusion score takes three indicators, namely structural similarity, edge continuity, and local contrast, as inputs, generates an overall quality evaluation value for the colony image block based on a multi-factor weighting mechanism, and feeds it back to the parameter adjustment module. Each sub-indicator is assigned a fixed weight coefficient according to its importance in the task. A linear superposition strategy is adopted during the fusion process, and the scale unity between indicators is also considered. The calculation formula is as follows: ; In the formula, is the weight coefficient, satisfying .

[0042] S5. Use the structural frequency curvature coupling to map the thermal imaging image structure, calculate the local frequency response, regional curvature value, and information entropy of the image, and construct a non-linear coupling mapping function.

[0043] Furthermore, in step S5, as Figure 4 shown, the specific steps of the structural frequency curvature coupling mapping of the infrared thermal imaging image are as follows.

[0044] S51. Perform wavelet transform on the original infrared thermal imaging image, extract its sub-band responses in the horizontal, vertical, and diagonal directions, which respectively reflect the texture intensity changes of the infrared thermal imaging image in different directions. For each pixel position, count the response amplitudes in the three-direction sub-bands, and obtain its overall frequency response intensity through an amplitude combination strategy to calculate the local frequency response of the image , and the calculation formula for the high-frequency component of each pixel point through wavelet transform is: ; In the formula, are the wavelet coefficients of the infrared thermal imaging image in the horizontal, vertical, and diagonal directions respectively, and the initial values are 1, 1, and 2 respectively; S52. By obtaining the gradient values of the infrared thermal imaging image in the horizontal and vertical directions, further calculate its second-order derivatives, including the horizontal change rate, vertical change rate, and mixed change rate, and introduce a small regularization factor during the processing to calculate the local curvature value , and the calculation formula based on the second-order derivative is: ; In the formula, is the local curvature value at the th pixel, indicating the degree of curvature of the surface in the area where the pixel is located, used to detect edges and curved areas in the image, is the gradient of the infrared thermal imaging image in the x direction, that is, the partial derivative, is the gradient of the infrared thermal imaging image in the y direction, is the second-order derivative of the infrared thermal imaging image in the x direction, is the mixed second-order derivative of the infrared thermal imaging image in the xy direction, is the second-order derivative of the infrared thermal imaging image in the y direction, is a small constant, taking , used to avoid a zero denominator; S53. Establish a sliding window centered on each pixel, count the gray value distribution of all pixels within the window, calculate its probability distribution, construct an entropy value function based on the probability density, and add a minimum bias during the construction of the probability density to calculate the information entropy , and the calculation formula based on the local gray distribution is: ; In the formula, is the probability distribution of the gray value , is a constant, initially set to 1 to prevent zero in logarithmic calculations; S54. The generation of the enhancement weight is based on three characteristic parameters: structural frequency, local curvature, and information entropy, and is comprehensively fused through a non-linear coupling mechanism. First, perform scale normalization processing on the above three types of features, and then allocate control factors according to task requirements to adjust the influence degree of each parameter on the enhancement weight. Subsequently, allocate control factors according to task requirements to adjust the influence degree of each parameter on the enhancement weight, and generate enhancement parameters through a non-linear coupling mapping. The calculation formula is: ; In the formula, is the final enhancement weight.

[0045] Furthermore, in step S5, for the infrared thermal imaging image enhancement model, it is written based on the Python language, using the Pytorch framework. The optimizer uses stochastic gradient descent, the initial learning rate is set to 0.01, the momentum is set to 0.9, the weight decay is set to 0.0001, the training batch is set to 64, the initial number of training epochs is set to 100, the cross-entropy loss function is used to calculate the loss of each iteration, and the model parameters are optimized according to the loss. After training, the structural similarity index is used to evaluate the performance of the model, reflecting the ability of the model to retain details and suppress noise during the enhancement process.

[0046] S6. Merge each infrared thermal imaging image block to obtain an overall enhanced infrared thermal imaging image output.

[0047] Further, in step S6, as Figure 5 shown, Figure 5 shows the effect diagram before the infrared thermal imaging image enhancement, and as Figure 6 shown, Figure 6 shows the enhanced image effect diagram obtained after processing by the infrared thermal imaging image enhancement model.

[0048] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the inventive concept of the present invention, several modifications and improvements can be made, and these all belong to the protection scope of the present invention.

Claims

1. An infrared thermal imaging image processing method for pain areas, characterized in that The following steps are involved: S1, construct the initial structure description, collect infrared thermal imaging images of the pain area, construct a preliminary abnormal area distribution map based on edge segmentation, and make a structure vector map; S2. According to the structure vector graph, the number, position, size and edge blur degree information of the abnormal area are extracted to form a structure guidance graph; S3, cutting the infrared thermal imaging image into multiple image blocks, and enhancing each image block according to its local features and structure guidance map; S4, calculating the three indicators of structural similarity, edge continuity and local contrast for the enhanced image block, calculating the quality score, and adjusting the enhancement parameters of the next image block according to the score; S5. Use structural frequency curvature coupling to map the thermal imaging image structure, calculate the image local frequency response, regional curvature value, information entropy, and construct a nonlinear coupling mapping function; S6. Merge each image block to obtain an overall enhanced thermal imaging image output.

2. The infrared thermal imaging image processing method for pain areas according to claim 1, characterized in that In the step S2, the method for constructing the structure guidance diagram specifically includes the following steps: extracting the spatial position information, scale information, and boundary blurriness index of each abnormal area from the abnormal area distribution diagram, and encoding the multi-dimensional attribute information into a feature combination for structure modeling. Among them, the description unit of each abnormal area includes coordinate information for indicating its position relationship in the infrared thermal imaging image, a scale parameter reflecting its growth range, and an edge blurriness parameter for measuring the clarity of the edge. The structure vector set is used to construct the overall structure guidance diagram .

3. The infrared thermal imaging image processing method for pain area according to claim 2, characterized in that In step S3, the image block cutting enhancement processing method is: S31. Based on a preset two-dimensional grid division strategy, the input infrared thermal imaging image is evenly divided in the spatial dimension to form a plurality of non-overlapping sub-image areas, each of which has the same size in the row and column directions, and the division result covers the entire infrared thermal imaging image. A clear position index mark is set for each image block, and region-level matching and correspondence are performed with the structure guidance map. S32. The enhancement processing method adopts an image block enhancement strategy based on feature response guidance. By analyzing the brightness distribution characteristics, structural guidance information and spatial position relationship of the current image block in the overall infrared thermal imaging image, the contrast adjustment amplitude and detail retention intensity of the image block are dynamically adjusted. The enhancement parameters are determined jointly based on the image statistical characteristics of the current block and the corresponding structural guidance area. Differentiated enhancement operations are applied to each image block in a block-by-block processing manner.

4. The infrared thermal imaging image processing method for pain area according to claim 3, wherein In step S4, the quality score is calculated as follows: S41, calculate the structural similarity SSIM of each image block after enhancement, by calculating the grayscale mean of the image blocks before and after enhancement and , grayscale variance and covariance between them Perform statistical extraction and build a comprehensive evaluation model based on brightness consistency, contrast preservation and structural reconstruction degree. After normalization and suppression term adjustment, generate an evaluation value representing the consistency of image block structure. S42, the edge continuity index is constructed by analyzing the difference in edge gradients before and after image block enhancement. First, the gradient amplitude map is extracted in each image block, and the gradient response values of each pixel position in the original image and the enhanced image are recorded respectively. Then, the point-by-point difference calculation of the response values before and after the same position is performed, and the average of the entire infrared thermal imaging image area is obtained to obtain the edge disturbance statistics, and the edge continuity after enhancement is calculated. The calculation formula is: ； Wherein, is the edge continuity score of the image block, is the th pixel gradient of the image block, is the th pixel gradient of the enhanced image block, is the number of pixels calculated within the image block; S43, calculate the brightness deviation value between each pixel in the image block and its neighborhood, and then normalize it into a standard contrast response, which is determined by the pixel value and the grayscale mean and standard deviation of the image block in which it is located. After performing the above processing on the image blocks before and after enhancement, calculate the difference between the two, and perform statistical aggregation in the entire block area to calculate the local contrast improvement score. The calculation formula is: ； Wherein, is the local contrast enhancement score of the image block, is the gray value of the th pixel of the image block, is the gray value of the th pixel of the enhanced image block, is the mean value of all pixels in the image block , is the mean value of all pixels in the enhanced image block ; S44. The fusion score takes structural similarity, edge continuity and local contrast as input, generates an overall image block quality evaluation value based on a multi-factor weighted mechanism, and feeds it back to the parameter adjustment module. Each sub-indicator is assigned a fixed weight coefficient according to its importance in the task. A linear superposition strategy is adopted in the fusion process, while considering the scale uniformity between indicators.

5. The infrared thermal imaging image processing method for pain area according to claim 4, wherein In the step S5, the calculation method of the coupled mapping of the infrared thermal imaging image structure frequency curvature is as follows: S51. Perform wavelet transform on the original infrared thermal imaging image, extract the sub-band responses in the horizontal, vertical, and diagonal directions, which respectively reflect the texture intensity changes of the infrared thermal imaging image in different directions. For each pixel position, statistically calculate the response amplitudes in the three-direction sub-bands, obtain the overall frequency response intensity through an amplitude combination strategy, and calculate the local frequency response of the image ; S52. By obtaining the gradient values of the infrared thermal imaging image in the horizontal and vertical directions, further calculating its second-order derivatives, including the horizontal change rate, the vertical change rate, and the mixed change rate, and introducing a small regularization factor during the processing to calculate the local curvature value ; S53. Establish a sliding window centered on each pixel, count the gray value distribution of all pixels within the window, calculate its probability distribution, construct an entropy value function based on the probability density, and add a very small bias to calculate the information entropy during the construction of the probability density ; S54. The generation of the enhanced weight is based on three characteristic parameters of structural frequency, local curvature and information entropy, and is comprehensively fused through a non-linear coupling mechanism. First, the above three types of features are subjected to scale normalization processing, and then control factors are allocated according to task requirements to adjust the influence degree of each parameter on the enhanced weight. Subsequently, control factors are allocated according to task requirements to adjust the influence degree of each parameter on the enhanced weight, and enhanced parameters are generated through non-linear coupling mapping.

Images