Region identification method and device of graphene thermal therapy robot

Abstract

The invention is suitable for the technical field of image processing, and provides a region identification method and device for a graphene thermal therapy robot, and the method comprises the steps: collecting an original image on a thermal therapy platform, and extracting a human body region in the original image; extracting a target to-be-recognized image corresponding to the target thermal therapy trunk part, and inputting the target to-be-recognized image into the trunk probability recognition model to obtain a probability distribution diagram output by the trunk probability recognition model; according to probability value distribution in the probability distribution diagram, extracting a target image region corresponding to the target thermal therapy trunk part in the target image to be recognized; and controlling the mechanical arm to move to an actual area corresponding to the target image area according to a conversion relation between the image coordinate system and the actual coordinate system to execute thermal therapy operation. The area identification method can adapt to different human body types and postures, and the identification process does not depend on manual intervention.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing, and in particular relates to a region recognition method and device for a graphene thermal therapy robot. Background Art

[0002] With the development of medical and health technologies, hyperthermia, as a non-invasive physical therapy, has been widely used in rehabilitation medicine, pain relief, and other fields. Traditional hyperthermia devices often rely on manual operation to apply heat or apply heat to a patient's body. This method has significant limitations in terms of precision, repeatability, and intelligence.

[0003] In recent years, with the continuous development of artificial intelligence, computer vision, and emerging materials (such as graphene), intelligent devices with automatic recognition and thermal therapy capabilities have gradually become a research hotspot. Graphene, due to its excellent thermal conductivity and flexible controllable properties, is widely used in the new generation of thermal therapy equipment. As intelligent medical devices that combine graphene thermal therapy with robotics, graphene thermal therapy robots offer advantages such as programmable control, automatic positioning, and personalized treatment, making them an important means of improving thermal therapy efficiency and treatment experience.

[0004] However, prior art graphene thermotherapy robots typically rely on manual calibration or simple image processing algorithms to identify the desired area on the human body before performing thermotherapy. These methods are not only inaccurate and significantly affected by posture and lighting variations, but also struggle to adapt to different body types and accurately locate specific areas. Furthermore, different areas of the body require different thermotherapy methods and intensities. Failure to accurately identify the target area can easily lead to poor thermotherapy results or treatment errors.

[0005] Therefore, there is an urgent need for a region recognition method based on image recognition that can automatically, efficiently and accurately identify the target hyperthermia therapy torso parts of the human body without relying on human intervention, and combine it with the control mechanism of the graphene hyperthermia therapy robot to realize intelligent and personalized hyperthermia therapy operations to improve the safety, comfort and intelligence level of treatment. Summary of the Invention

[0006] In view of this, an embodiment of the present invention provides a region recognition method and device for a graphene thermal therapy robot to solve the technical problem that traditional technologies are difficult to achieve adaptability between different human bodies and precise positioning of torso parts.

[0007] A first aspect of an embodiment of the present invention provides a region recognition method for a graphene thermotherapy robot. The region recognition method for the graphene thermotherapy robot is applied to the graphene thermotherapy robot. The graphene thermotherapy robot includes a robotic arm, a graphene thermotherapy module, a visual acquisition module, and a control module. The region recognition method for the graphene thermotherapy robot includes:

[0008] collecting an original image on the thermal therapy platform and extracting a human body region in the original image;

[0009] Based on a preset human body torso ratio, the human body region is segmented to obtain images to be identified corresponding to a plurality of torso parts;

[0010] Extracting a target image to be identified corresponding to a target hyperthermia torso part, and inputting the target image to be identified into a torso probability recognition model to obtain a probability distribution map output by the torso probability recognition model; the probability distribution map refers to a probability data matrix composed of torso probabilities corresponding to a plurality of pixel points, wherein the torso probability refers to the probability that the current pixel point is the pixel point corresponding to the target hyperthermia torso part; the target hyperthermia torso part refers to the torso part on which a hyperthermia operation is to be performed, and different hyperthermia torso parts correspond to different torso probability recognition models;

[0011] extracting a target image region corresponding to a target hyperthermia treatment torso part in the target image to be identified according to the probability value distribution in the probability distribution graph;

[0012] According to the conversion relationship between the image coordinate system and the actual coordinate system, the robotic arm is controlled to move to the actual area corresponding to the target image area to perform the thermal treatment operation.

[0013] Furthermore, the step of dividing the human body region into multiple images to be identified corresponding to the torso parts based on the preset human body torso proportions includes:

[0014] Based on the first length direction of the human body region, dividing the human body region according to the preset human body torso ratio to obtain a plurality of first image regions;

[0015] respectively acquiring first extended areas of the first image area in the length direction;

[0016] splicing the first image area with a first extended area corresponding to the first image area to obtain a redundant image area;

[0017] Inputting the redundant image area corresponding to the head-to-torso ratio into a head recognition model to obtain a recognition result output by the head recognition model;

[0018] If the recognition result is the head area, the redundant image area is used as the image to be recognized;

[0019] If the recognition result is a non-head region, segmenting the human body region according to the preset human body torso ratio based on the second length direction of the human body region to obtain a plurality of second image regions;

[0020] respectively acquiring second extended areas of the second image area in the length direction;

[0021] The second image area is spliced with a second extended area corresponding to the second image area to obtain the image to be recognized.

[0022] Furthermore, the step of extracting the target image to be identified corresponding to the target hyperthermia treatment torso part, inputting the target image to be identified into the torso probability recognition model, and obtaining the probability distribution map output by the torso probability recognition model includes:

[0023] Scaling the target image to be identified to obtain target images to be identified at multiple scales;

[0024] Inputting the target image to be identified into a first convolutional network to obtain first feature data output by the first convolutional network; the first feature data is used to characterize common features in the target image to be identified;

[0025] The first feature data is input into a second convolutional network to obtain a probability distribution map output by the second convolutional network.

[0026] Furthermore, the step of inputting the first feature data into a second convolutional network to obtain a probability distribution map output by the second convolutional network includes:

[0027] Inputting the first feature data into a plurality of sub-convolutional networks respectively to obtain a plurality of second feature data respectively output by the plurality of sub-convolutional networks; the plurality of sub-convolutional networks respectively use convolution kernels of different scales;

[0028] Inputting the plurality of second feature data into a fully connected layer respectively to obtain a plurality of classification probabilities output by the fully connected layer; the fully connected layer is used to classify the target hyperthermia treatment torso part;

[0029] Normalize multiple classification probabilities to obtain multiple target probabilities;

[0030] Multiply each second feature data and the target probability corresponding to the second feature data to obtain third feature data;

[0031] The plurality of third feature data are superimposed to obtain the probability distribution graph.

[0032] Furthermore, the step of superimposing a plurality of third feature data to obtain the probability distribution graph includes:

[0033] Adding characteristic values of the same pixel position in the plurality of third characteristic data to obtain fourth characteristic data;

[0034] extracting a maximum value from the fourth feature data;

[0035] The probability distribution graph is obtained by dividing the characteristic value corresponding to each pixel position in the fourth characteristic data by the maximum value.

[0036] Furthermore, the step of extracting a target image region corresponding to a target hyperthermia treatment torso part in the target image to be identified according to the probability value distribution in the probability distribution graph includes:

[0037] Extract the probability distribution maps corresponding to multiple scales;

[0038] Interpolating the probability distribution maps corresponding to the multiple scaling scales to obtain multiple interpolated probability distribution maps; the size of the interpolated probability distribution map is consistent with the original size of the target image to be recognized; the scaling scale refers to the size obtained by scaling the original size;

[0039] Extracting a first pixel point whose probability value is greater than a first threshold value in the probability distribution map corresponding to the original size;

[0040] Extracting a second pixel point having a probability value greater than a first threshold value from the interpolation probability distribution map;

[0041] A target image area corresponding to a target hyperthermia treatment torso part in the target image to be identified is extracted according to the coordinate data of the first pixel point and the second pixel point.

[0042] Furthermore, the step of extracting a target image area corresponding to a target hyperthermia treatment torso part in the target image to be identified based on the coordinate data of the first pixel point and the second pixel point includes:

[0043] Extracting image coordinate data corresponding to all first pixel points and all second pixel points;

[0044] Counting the number of the first pixel point or the second pixel point corresponding to each of the image coordinate data; if the number of the pixel points corresponding to the current image coordinate data is greater than a preset number, using the pixel point corresponding to the current image coordinate data as the pixel point corresponding to the target hyperthermia treatment torso part;

[0045] The multiple pixel points corresponding to the target hyperthermia treatment torso area are used as the target image area.

[0046] A second aspect of an embodiment of the present invention provides a region recognition device for a graphene hyperthermia therapy robot, comprising:

[0047] An acquisition unit, configured to acquire an original image on the thermal therapy platform and extract a human body region from the original image;

[0048] a segmentation unit, configured to segment the human body region based on a preset human body torso ratio to obtain images to be identified corresponding to a plurality of torso parts;

[0049] an input unit for extracting a target image to be identified corresponding to a target hyperthermia torso part, and inputting the target image to be identified into a torso probability recognition model to obtain a probability distribution map output by the torso probability recognition model; the probability distribution map is a probability data matrix composed of torso probabilities corresponding to a plurality of pixel points, the torso probability being the probability that a current pixel point is a pixel point corresponding to a target hyperthermia torso part; the target hyperthermia torso part being the torso part requiring hyperthermia treatment, and different hyperthermia torso parts corresponding to different torso probability recognition models;

[0050] An extraction unit, configured to extract a target image region corresponding to a target hyperthermia treatment torso part in the target image to be identified according to a probability value distribution in the probability distribution map;

[0051] The control unit is used to control the robot arm to move to the actual area corresponding to the target image area to perform the thermal treatment operation according to the conversion relationship between the image coordinate system and the actual coordinate system.

[0052] A third aspect of an embodiment of the present invention provides a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the region recognition method of the graphene thermal therapy robot described in the first aspect are implemented.

[0053] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the computer program implements the steps in the region recognition method of the graphene thermal therapy robot described in the first aspect.

[0054] Compared to the prior art, the present invention offers the following advantages: by capturing raw images from a thermotherapy platform and segmenting them based on preset human torso proportions, the effects of external factors such as posture and lighting are effectively reduced, enabling precise delineation and identification of human thermotherapy regions. The present invention employs a thermotherapy region recognition method based on a probabilistic torso recognition model, enabling accurate identification of different torso regions. By selecting corresponding recognition models for different target torso regions, the method ensures accurate thermotherapy treatment for each region and avoids adverse treatment outcomes caused by recognition errors or inaccuracies. This method converts the target image region into an actual coordinate system, enabling automatic positioning and precise control of the robotic arm. Based on the recognition results, the robotic arm automatically moves to the target thermotherapy region and implements the corresponding thermotherapy operation, improving treatment efficiency and enabling personalized adjustments based on the patient's specific needs, enhancing treatment comfort. The region recognition method of the present invention adapts to different body shapes and postures, and the recognition process does not rely on manual intervention. Therefore, it can be widely applied to different individuals and scenarios, demonstrating strong adaptability and universality. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0056] Figure 1 A schematic flow chart of a region recognition method for a graphene hyperthermia therapy robot provided by the present invention is shown;

[0057] Figure 2 A schematic diagram of a region recognition device of a graphene hyperthermia robot provided by one embodiment of the present invention is shown;

[0058] Figure 3 A schematic diagram of a terminal device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0059] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

[0060] The embodiments of the present invention provide a region recognition method and device for a graphene hyperthermia robot to solve the technical problems of relying on traditional physical methods or artificial vision to judge the abnormal state of a display device, resulting in low detection accuracy and poor real-time performance.

[0061] First, the present invention provides a region recognition method for a graphene hyperthermia robot. Figure 1 , Figure 1 FIG1 shows a schematic flow chart of a region recognition method of a graphene hyperthermia robot provided by the present invention. Figure 1 As shown, the region recognition method of the graphene hyperthermia therapy robot may include the following steps:

[0062] Step 101: collecting an original image on a thermal therapy platform and extracting a human body region in the original image;

[0063] The robot uses a visual acquisition module (e.g., a camera) to capture raw images of a human body on the thermal therapy platform. These images contain the body's overall structure, and the system uses this image data to identify and locate areas requiring thermal therapy. Using image processing techniques, the robot extracts the human body region from the raw images. Human body image extraction techniques include, but are not limited to, edge detection, threshold processing, or deep learning models to identify the human body's outline in the image. The human body region can be obtained by removing pixels on the thermal therapy platform from the raw images.

[0064] Step 102: based on a preset human body torso ratio, the human body region is segmented to obtain images to be identified corresponding to a plurality of torso parts;

[0065] Based on known human anatomical proportions, the robot uses preset torso proportions (such as the size ratios of various torso parts) to segment the entire human body into multiple distinct torso parts. Each segmented area corresponds to a separate image to be identified, facilitating subsequent recognition and analysis. Torso parts include, but are not limited to, the head and torso, the upper torso, and the lower torso. The upper torso includes, but is not limited to, the neck, shoulders, arms, hands, chest, abdomen, or back, while the lower torso includes, but is not limited to, the buttocks, thighs, knees, calves, and feet.

[0066] Specifically, step 102 includes steps 1021 to 1028:

[0067] Step 1021: based on the first length direction of the human body region, dividing the human body region according to the preset human body torso ratio to obtain a plurality of first image regions;

[0068] The human body region is segmented along a specific "length direction" (e.g., the longitudinal direction from head to toe). This segmentation is determined based on preset torso proportions, with different parts divided based on standard proportions of human anatomy (e.g., the longitudinal length of the head, chest, abdomen, or legs). For example, if the preset torso proportions indicate that the body region should be segmented into head, chest, abdomen, etc., the system will segment the human body region into multiple "first image regions" along the length direction according to these proportions.

[0069] Step 1022: Acquire first extended areas of the first image area in the length direction respectively;

[0070] For each segmented first image region (e.g., chest, abdomen), an additional extended region is captured along its length. This extended region can help add context to the image during recognition, avoiding omissions or recognition errors caused by overly precise segmentation.

[0071] Step 1023: splicing the first image area with a first extended area corresponding to the first image area to obtain a redundant image area;

[0072] The first image region and its corresponding extended region are concatenated to form a "redundant image region." This redundant region includes the first image region itself and its extended portion, providing more contextual data for subsequent recognition models, helping to improve recognition accuracy.

[0073] Step 1024: Input the redundant image area corresponding to the head-to-torso ratio into the head recognition model to obtain a recognition result output by the head recognition model;

[0074] The spliced redundant image regions are fed into a pre-trained head recognition model. This model's task is to analyze the image and determine whether it contains a head region. The model outputs a recognition result indicating whether the redundant image region contains a head region.

[0075] Step 1025: If the recognition result is the head region, the redundant image region is used as the image to be recognized;

[0076] If the recognition result indicates that the redundant image region does contain the head region, then the redundant image region is regarded as an image to be recognized and is used for subsequent thermal therapy operations.

[0077] Step 1026: If the recognition result is a non-head region, segmenting the human body region according to the preset human body torso ratio based on the second length direction of the human body region to obtain a plurality of second image regions;

[0078] If the recognition result is a non-head area (such as a foot or a leg), the subsequent processing will use the "second length direction" of the human body area for segmentation to correct the head area.

[0079] It is understandable that, since the user's lying direction on the thermal therapy platform may deviate, the head position needs to be verified to perform segmentation in the correct direction.

[0080] Step 1027: Acquire second extended areas of the second image area in the length direction respectively;

[0081] For each second image region, an extended region is obtained along its length. This is done to ensure that the context around the target region is covered during recognition, avoiding inaccurate recognition due to the limitations of cutting.

[0082] Step 1028: Splice the second image area with the second extended area corresponding to the second image area to obtain the image to be recognized.

[0083] The second image region is joined with the corresponding second extended region to obtain a complete image region to be recognized. This region will be used as the new input image for subsequent processing, recognition, and analysis.

[0084] In the embodiments corresponding to steps 1021 through 1028, a flexible image segmentation and splicing strategy dynamically adjusts the image segmentation method based on the recognition results, thereby ensuring accurate recognition. By introducing redundant image regions, the system can provide more contextual information during recognition, thereby improving the accuracy of identifying the head or other body parts. If initial head recognition fails, the system automatically switches to identifying other body parts, ensuring the stability and comprehensiveness of the entire process.

[0085] Step 103: Extracting a target image to be identified corresponding to the target hyperthermia torso part, and inputting the target image to be identified into a torso probability recognition model to obtain a probability distribution graph output by the torso probability recognition model; the probability distribution graph is a probability data matrix composed of torso probabilities corresponding to a plurality of pixel points, wherein the torso probability is the probability that the current pixel point is the pixel point corresponding to the target hyperthermia torso part; the target hyperthermia torso part is the torso part requiring hyperthermia treatment, and different hyperthermia torso parts correspond to different torso probability recognition models;

[0086] Among these segmented images, the system will select an image to be identified corresponding to the target torso part (i.e., the target image to be identified). The image of this part is extracted and passed to the recognition model as input data. The selected target image to be identified is input into a trained torso probability recognition model. The task of this model is to analyze the input image and predict the probability that each pixel in the image belongs to the target hyperthermia part. These probabilities will help identify the precise area for the hyperthermia operation. The torso probability recognition model outputs a probability distribution map, which is a probability data matrix in which each pixel has a corresponding probability value, indicating the possibility that the pixel belongs to the target hyperthermia part. Areas with high probabilities correspond to hyperthermia parts, and areas with low probabilities indicate that they are not within the hyperthermia range.

[0087] Specifically, step 103 includes steps 1031 to 1033:

[0088] Step 1031: scaling the target image to be identified to obtain target images to be identified at multiple scales;

[0089] The target image to be identified is scaled, meaning the original image is scaled multiple times to different scales. This allows the model to learn detailed information from images at different scales. Images of different scales help capture features of different sizes and resolutions, which is particularly important for tasks such as torso recognition, as different parts of the body may appear at different scales in the image. For example, the head may be smaller, while the chest may be larger. Scaling multiple scaled images helps enhance the robustness of the model, allowing it to correctly identify target parts at different scales.

[0090] Step 1032: Inputting the target image to be identified into a first convolutional network to obtain first feature data output by the first convolutional network; the first feature data is used to characterize common features in the target image to be identified;

[0091] The first convolutional network (usually a deep convolutional neural network) processes the input multi-scale image and extracts image features through a series of convolution operations. Convolutional neural networks excel at extracting spatial features from images, capturing local structures such as edges and textures. The output "first feature data" are the image features extracted by the convolution operation. They are used to represent common features in the image, such as shape, edges, texture, etc., or salient features of different parts of the image.

[0092] The first feature data represents the "common features" of the target image to be identified. These common features are commonly found in the image and are recognizable, such as the outlines of various human body parts. The feature data extracted by the convolutional network provides the basis for subsequent target region identification.

[0093] Step 1033: Input the first feature data into a second convolutional network to obtain a probability distribution map output by the second convolutional network.

[0094] The second convolutional network receives the feature data extracted by the first convolutional network and further processes these features. In this step, the second convolutional network is usually used to further perform deep learning on the extracted features, gradually increasing the level of abstraction of the features and predicting the target area.

[0095] The output "probability distribution map" is a probability matrix that represents the probability that each pixel in the image belongs to a different body part. The probability value of each pixel indicates the likelihood that the location is part of the target body part.

[0096] For example, when identifying the abdomen area, a portion of the area in the corresponding probability distribution map will have a higher probability value, indicating the abdomen area, while the probability values of other areas will be lower.

[0097] In the embodiment corresponding to steps 1031 to 1033, the target image to be identified is processed at multiple scales, followed by step-by-step analysis and feature extraction using two convolutional networks, ultimately generating a probability distribution map for the torso region. This process leverages the powerful feature extraction capabilities of convolutional neural networks, enabling the model to accurately identify different parts of the human torso at different scales. The first convolutional network extracts the basic features of the image, while the second convolutional network generates a detailed probability distribution map based on these features, helping the robot accurately locate the torso area requiring thermal therapy.

[0098] Specifically, step 1033 includes steps A1 to A5:

[0099] Step A1: inputting the first feature data into a plurality of sub-convolutional networks respectively to obtain a plurality of second feature data respectively output by the plurality of sub-convolutional networks; the plurality of sub-convolutional networks respectively use convolution kernels of different scales;

[0100] In this process, the feature data extracted by the first convolutional network (i.e., the first feature data) is first input into multiple sub-convolutional networks. Each sub-convolutional network uses a convolution kernel of a different scale. Each sub-convolutional network extracts specific features of the area of interest, such as shape, edges, and texture, thereby generating multiple "second feature data."

[0101] Convolution kernels of different scales can capture features of different scales in the image. Small-scale convolution kernels (such as 3x3) are usually good at capturing local details and edge information, while large-scale convolution kernels (such as 7x7 or larger) can capture larger structures and global information. By combining convolution kernels of different scales, the network can obtain both fine-grained and coarse-grained features, enhancing its understanding of complex images. Using convolution kernels of different scales can enrich the network's feature expression capabilities. The features extracted by each convolution kernel are an expression of the original image from a different perspective. These diverse features are combined to provide a more comprehensive and detailed image description, which helps improve the classification and recognition performance of the network. Multi-scale feature extraction helps improve the generalization ability of the network, allowing it to perform well when processing unseen images. This is because convolution kernels of different scales can learn multiple patterns and structures of the image, reducing the risk of overfitting.

[0102] Step A2: Inputting the plurality of second feature data into a fully connected layer respectively to obtain a plurality of classification probabilities output by the fully connected layer; the fully connected layer is used to classify the target hyperthermia treatment torso part;

[0103] In this step, the second feature data from each sub-convolutional network is input to the fully connected layer. The fully connected layer (for classification) uses this feature data to determine the probability that each image region contains the target hyperthermia treatment torso region. The fully connected layer performs a weighted summation and nonlinear activation on the input second feature data, outputting the classification probability corresponding to the current second feature data. These probabilities represent the likelihood that the target hyperthermia treatment region appears in that image region.

[0104] Step A3: Normalize multiple classification probabilities to obtain multiple target probabilities;

[0105] Normalization is the process of adjusting different classification probabilities to the same standard range (usually between 0 and 1) for comparison and further processing. Softmax function is usually used for normalization, which converts each classification probability into a proportion value, ensuring that the sum of all probabilities is 1.

[0106] Step A4: multiply each second feature data by the target probability corresponding to the second feature data to obtain third feature data;

[0107] The purpose of this process is to weight each second feature data with its corresponding target probability. By multiplying the feature data with the probability, the system can strengthen its focus on the target hyperthermia treatment torso area.

[0108] It is understandable that since different sub-convolutional networks use different convolution kernels, different convolution kernels have different feature extraction capabilities for different torso areas. In order to integrate the feature extraction capabilities of multiple convolution kernels, the second feature data and the target probability corresponding to the second feature data are multiplied to adjust the proportion of feature data corresponding to different convolution kernels in the comprehensive data, thereby improving the feature representation ability of the feature data.

[0109] Step A5: Superimpose multiple third feature data to obtain the probability distribution diagram.

[0110] During this process, all weighted third feature data are superimposed to form the final output—a probability distribution map. This superimposed third feature data provides a comprehensive probability distribution for each image region, indicating the comprehensive probability that each pixel in the image belongs to the target hyperthermia treatment area. The resulting probability distribution map provides accurate region identification for the hyperthermia treatment system.

[0111] In the embodiments corresponding to steps A1 through A5, multiple sub-convolutional networks employ convolution kernels of varying scales, followed by classification via a fully connected layer, ultimately yielding a comprehensive probability distribution map. This is then weighted by target probability to enhance focus on key areas. This approach enables the system to accurately identify and locate target hyperthermia treatment areas, providing efficient and personalized treatment plans.

[0112] Specifically, step A5 includes steps A51 to A53:

[0113] Step A51: adding the feature values of the same pixel position in the plurality of third feature data to obtain fourth feature data;

[0114] In the previous step, multiple "third feature data" were obtained by multiplying the second feature data with its corresponding target probability. These third feature data represent the weighted features corresponding to different convolution kernels. The goal of this step is to superimpose multiple third feature data at each pixel position, that is, for each pixel position, add up the values of all third feature data. This is equivalent to summing up all the weighted features of each pixel point, so that the final feature value of each pixel point represents the comprehensive information from all parts.

[0115] Step A52: extracting the maximum value of the fourth characteristic data;

[0116] This step further processes the fourth eigenvalues to find their maximum value. This step is typically done to emphasize the most important features and ensure that the maximum value dominates the subsequent processing. By extracting the maximum value, it can be used to normalize the fourth eigenvalues so that the value of each pixel position is within a standard range (usually between 0 and 1). This maximum value typically represents the most significant feature in the image, and the other eigenvalues are scaled accordingly.

[0117] Step A53: Divide the characteristic value corresponding to each pixel position in the fourth characteristic data by the maximum value to obtain the probability distribution map.

[0118] This step normalizes the fourth feature data by dividing the value of each pixel by the maximum value in the fourth feature data. In this way, the values of all pixel locations will be compressed into a standard range (between 0 and 1). Normalization is to ensure that the magnitude of each pixel value in the probability distribution map is between 0 and 1, so that they can be used as probabilities. Maximum value normalization helps to highlight relatively large feature values while maintaining the proportional relationship of other features.

[0119] In the embodiment corresponding to steps A51 to A53, a "fourth feature data" is obtained by adding multiple third feature data at each pixel position, which summarizes the feature information extracted from different convolution kernels. By extracting the maximum value in the fourth feature data, the most significant feature value in the image is found. This helps to control the normalization range in subsequent processing to ensure that the significant features in the result are emphasized. Each pixel value in the fourth feature data is divided by the maximum value to normalize all feature values to between 0 and 1, generating a final probability distribution map. This step ensures that the generated probability distribution map can be used as a valid probability map for identifying the thermal therapy torso part, and all values are within the standardized range. The final probability distribution map will represent the probability that each pixel position in the image belongs to a specific torso part. This map can be used for further decision-making or processing, such as determining the thermal therapy site.

[0120] Step 104: extracting a target image region corresponding to a target hyperthermia treatment torso part in the target image to be identified according to the probability value distribution in the probability distribution graph;

[0121] By analyzing the probability distribution graph, the system can extract regions with higher probability values from the image to be identified. These regions are considered to be the locations of the target torso parts for hyperthermia treatment. This target image region is the part that needs hyperthermia treatment.

[0122] Specifically, step 104 includes steps 1041 to 1045:

[0123] Step 1041: extracting probability distribution maps corresponding to multiple scales;

[0124] By scaling the original image, we generate multiple images of different sizes. Each scale is processed by the model to generate a probability distribution map. These probability distribution maps at different scales can reflect the likelihood of the target at different sizes, providing more comprehensive information to help locate the target area.

[0125] Step 1042: interpolating the probability distribution maps corresponding to the multiple scaling scales to obtain multiple interpolated probability distribution maps; the size of the interpolated probability distribution maps is consistent with the original size of the target image to be recognized; the scaling scale refers to the size obtained by scaling the original size;

[0126] Since the probability distribution maps of multiple scales are of different sizes, they need to be adjusted to the same size for unified processing. For the probability distribution map of each scale, an interpolation method is used to adjust it to the same size as the original image. This can be achieved through common image interpolation methods (such as bilinear interpolation, cubic interpolation, etc.). For the probability distribution map at each scaling scale, interpolation is used to adjust it to the same size as the original image, ensuring that the final interpolated probability distribution map matches the size of the target image to be identified. Through interpolation, it is ensured that the probability distribution maps obtained at different scales are consistent with the size of the original image, so that features at different scales can be directly compared.

[0127] Step 1043: extracting the first pixel point whose probability value is greater than a first threshold value in the probability distribution map corresponding to the original size;

[0128] The original-size probability distribution map reflects the likelihood of each pixel in the image corresponding to the target hyperthermia treatment area. By setting a threshold, regions with higher probabilities can be extracted to help locate potential hyperthermia treatment areas. Within the original-size probability distribution map, all pixels with probabilities greater than a preset "first threshold" are extracted. These probabilities represent the system's highest confidence level in the hyperthermia treatment area.

[0129] For example, if the first threshold is set to 0.8, then all pixels with a probability greater than 0.8 are candidate areas for the target hyperthermia treatment area. By setting the threshold, the pixels most likely to belong to the hyperthermia treatment area are screened out, and these pixels will become the basis for subsequent processing.

[0130] Step 1044: extracting a second pixel point having a probability value greater than a first threshold value from the interpolation probability distribution map;

[0131] After the interpolation process, the interpolated probability distribution map is consistent with the size of the original image. At this time, it is also necessary to extract the corresponding pixel points from the interpolated probability distribution map, especially those areas with higher probability values. Similar to the previous operation in the original size image, it is necessary to extract all pixel points with probability values greater than the first threshold in the interpolated probability distribution map. The probability values of these points represent the hyperthermia areas inferred at different scales. Similarly, a threshold is set (such as 0.8), and all pixel points with probability values higher than the threshold in the interpolated map are extracted. Extracting high-probability pixel points in the interpolated probability distribution map provides more information for the final determination of the target area, especially the potential location of the hyperthermia area at different scales.

[0132] Step 1045: extracting a target image region corresponding to the target thermal therapy torso part in the target image to be identified based on the coordinate data of the first pixel point and the second pixel point.

[0133] The first pixel point and the second pixel point correspond to the pixel positions with higher probability values in the original image and the interpolated image, respectively. By comparing the pixel positions at these two different scales, the target area can be determined more accurately. According to the coordinate data of the first pixel point (high probability point from the original image) and the second pixel point (high probability point from the interpolated image), the image area of the target hyperthermia torso part is determined. The key to this step is to compare and integrate the coordinates of the high probability pixel points at different scales to obtain the final target area. According to the multi-scale probability distribution map, combined with the information of the high probability pixel points, the image area of the target hyperthermia torso part is accurately extracted. The goal of this process is to generate a more accurate target area as the basis for subsequent hyperthermia treatment.

[0134] In the embodiment corresponding to steps 1041 to 1045, a multi-scale strategy is used to extract probability distribution maps of the target hyperthermia treatment torso area at different scales and resize them to a uniform size through interpolation. By setting a threshold, pixels with high probability values are extracted from the original image and the interpolated image, screening candidate regions for hyperthermia treatment. By comparing the coordinates of high-probability pixels in the original image and the interpolated image, and combining information from different scales, the target region corresponding to the target hyperthermia treatment torso area is accurately extracted. The resulting target image region can thus serve as the basis for identifying and processing the hyperthermia treatment torso area, thereby optimizing the location of the hyperthermia treatment region and the therapeutic effect.

[0135] Specifically, step 1045 includes steps B1 to B4:

[0136] Step B1: extracting image coordinate data corresponding to all first pixel points and all second pixel points;

[0137] In the previous steps, pixels with high probability values have been extracted from the original image (first pixel) and the interpolated image (second pixel). The coordinates of these pixel points represent the position of the target thermal therapy torso at different scales. First, it is necessary to extract the coordinate data corresponding to all first pixel points (pixels with probability values greater than the threshold in the original image) and all second pixel points (pixels with probability values greater than the threshold in the interpolated image). These coordinate data are usually the (x, y) coordinates of the image, representing the positions of these pixel points. For example, assuming that the probability values of some pixel points in the original image are greater than the threshold, write down the (x, y) coordinates of these points, and extract the pixel points in the interpolated image in the same way. By extracting these coordinate data, the position information of all high-probability areas in the image can be obtained, providing basic data for subsequent statistics and analysis.

[0138] Step B2: Counting the number of the first pixel point or the second pixel point corresponding to each of the image coordinate data;

[0139] After obtaining the coordinate data of all pixels, they need to be counted to identify which areas have a sufficient number of high-probability pixels. The more pixels in an area, the higher the probability that the area belongs to the hyperthermia treatment torso. Count the frequency of occurrence of each coordinate position, that is, which coordinate data appears most frequently in the first pixel and the second pixel. The goal of this step is to find out which pixel positions appear frequently in both the original image and the interpolated image. The pixel points at these positions may correspond to the main areas of the hyperthermia treatment torso. For example, if a certain coordinate point (x, y) appears more frequently in the first pixel and the second pixel, it means that the area is more likely to be the target hyperthermia treatment torso. By counting the number of coordinate points, it is possible to determine which areas have higher probability and consistency, thereby inferring the area of the target hyperthermia treatment torso.

[0140] Step B3: If the number of pixel points corresponding to the current image coordinate data is greater than a preset number, the pixel points corresponding to the current image coordinate data are used as pixel points corresponding to the target hyperthermia treatment torso part;

[0141] Not all high-probability pixels can form a significant hyperthermia area, and a "preset number" threshold needs to be set. This threshold is used to filter out those pixels that are not obvious or concentrated enough to ensure that the extracted area represents a significant hyperthermia torso part. For each image coordinate data, if the number of pixels corresponding to it exceeds the preset threshold (for example, exceeds a predetermined number of pixels), it is considered that the area to which the coordinate point belongs is significant enough to be part of the target hyperthermia torso part. By setting a threshold for the number of pixels, low-frequency or overly scattered pixels are filtered out to ensure that the final extracted area is a hyperthermia area with a certain degree of coherence and representativeness.

[0142] Step B4: multiple pixel points corresponding to the target hyperthermia treatment torso area are used as the target image area.

[0143] After determining which pixels belong to the torso area for hyperthermia treatment, the next step is to cluster these pixels to form a single target region. This region can be used for further positioning and processing of the hyperthermia treatment. The pixels filtered in the previous step (i.e., those with coordinates that meet the threshold requirements) are grouped into a region.

[0144] In the embodiment corresponding to steps B1 to B4, the coordinate data of high-probability pixel points are extracted from the original image and the interpolated image to represent the potential position of the target hyperthermia torso part. The number of occurrences of each coordinate point is counted to identify which positions have a higher frequency. By comparing with a preset number threshold, those pixel points with a higher frequency of occurrence are screened out and considered to correspond to the main area of the target hyperthermia torso part. The screened pixel points are combined into an overall area to clarify the position of the target hyperthermia torso part for subsequent processing. Through these steps, the target area of the hyperthermia torso part can be extracted from the probability distribution map, thereby providing accurate positioning information for the implementation of hyperthermia.

[0145] Step 105: According to the conversion relationship between the image coordinate system and the actual coordinate system, the robot arm is controlled to move to the actual area corresponding to the target image area to perform a thermal treatment operation.

[0146] By using the known transformation relationship between the image coordinate system and the real-world coordinate system, the system converts the target area in the image into coordinates in real space. The system then controls the robotic arm to move precisely according to these coordinates and perform the thermal therapy operation based on the graphene thermal therapy module.

[0147] Among them, the graphene hyperthermia module adopts graphene heating technology, which has a faster heating speed, more uniform heat distribution and stronger far-infrared radiation penetration compared to traditional electrothermal therapy, which can promote blood circulation in deep tissues and improve treatment efficiency. Equipped with an intelligent temperature control system, it monitors the temperature of the hyperthermia area in real time through temperature sensors and makes dynamic adjustments to prevent local overheating or burns, ensuring the safety of treatment. The intelligent adaptive adjustment system can adjust the intensity of hyperthermia based on the patient's physiological parameters (such as skin temperature, pain level, etc.) to make the treatment more in line with individual needs and improve patient comfort. It has a personalized treatment plan storage function, which can record the patient's previous hyperthermia data and combine it with AI optimization algorithms to provide long-term personalized health management plans. Compared with traditional hyperthermia equipment, it heats up faster, conducts heat more evenly, and has stronger far-infrared penetration. Combining temperature sensors and patient physiological data, it intelligently adjusts temperature and treatment time to improve safety.

[0148] In the embodiment corresponding to steps 101 to 105, by capturing the original image of the thermotherapy platform and segmenting it based on preset human torso proportions, the effects of external factors such as posture and lighting are effectively reduced, achieving precise demarcation and identification of human thermotherapy regions. The present invention utilizes a thermotherapy region recognition method based on a probabilistic torso recognition model, enabling accurate identification of different torso regions. By selecting corresponding recognition models for different target torso regions, this ensures accurate thermotherapy treatment for each region and avoids adverse treatment outcomes caused by recognition errors or inaccuracies. This method automatically positions and precisely controls the robotic arm by converting the target image region into the actual coordinate system. Based on the recognition results, the robotic arm automatically moves to the target thermotherapy region and implements the corresponding thermotherapy operation, improving treatment efficiency and enabling personalized adjustments based on the patient's specific needs, enhancing treatment comfort. The region recognition method of the present invention is adaptable to different body shapes and postures, and the recognition process does not rely on manual intervention. Therefore, it can be widely applied to different individuals and scenarios, demonstrating strong adaptability and universality.

[0149] like Figure 2 The present invention provides a region recognition device for a graphene hyperthermia robot, see Figure 2 , Figure 2 FIG. 1 shows a schematic diagram of a region recognition device of a graphene hyperthermia robot provided by the present invention, as shown in FIG. Figure 2 The region recognition device of a graphene hyperthermia therapy robot includes:

[0150] An acquisition unit 21 is used to acquire an original image on the thermal therapy platform and extract a human body region from the original image;

[0151] A segmentation unit 22 is configured to segment the human body region based on a preset human body torso ratio to obtain images to be identified corresponding to a plurality of torso parts;

[0152] An input unit 23 is configured to extract a target image to be identified corresponding to a target hyperthermia torso part, and input the target image to be identified into a torso probability recognition model to obtain a probability distribution map output by the torso probability recognition model; the probability distribution map is a probability data matrix composed of torso probabilities corresponding to a plurality of pixel points, wherein the torso probability is the probability that the current pixel point is the pixel point corresponding to the target hyperthermia torso part. The target hyperthermia torso part is the torso part requiring hyperthermia treatment, and different hyperthermia torso parts correspond to different torso probability recognition models.

[0153] An extraction unit 24 is configured to extract a target image region corresponding to a target hyperthermia treatment torso part in the target image to be identified according to the probability value distribution in the probability distribution map;

[0154] The control unit 25 is used to control the robot arm to move to the actual area corresponding to the target image area to perform the thermal treatment operation according to the conversion relationship between the image coordinate system and the actual coordinate system.

[0155] The present invention provides a region recognition device for a graphene thermotherapy robot. This device captures raw images from a thermotherapy platform and segments them based on preset human torso proportions, effectively reducing interference from external factors such as posture and lighting, enabling precise demarcation and identification of human thermotherapy regions. The present invention employs a thermotherapy region recognition method based on a probabilistic torso recognition model, enabling accurate identification of different torso regions. By selecting corresponding recognition models for different target torso regions, the device ensures accurate thermotherapy treatment for each region and avoids adverse treatment outcomes caused by recognition errors or inaccuracies. This method converts the target image region into an actual coordinate system, enabling automatic positioning and precise control of the robot arm. Based on the recognition results, the robot arm automatically moves to the target thermotherapy region and performs the corresponding thermotherapy operation, improving treatment efficiency and enabling personalized adjustments based on the patient's specific needs, enhancing treatment comfort. The region recognition method of the present invention adapts to different body shapes and postures, and the recognition process does not rely on manual intervention. Therefore, it can be widely applied to different individuals and scenarios, demonstrating strong adaptability and universality.

[0156] Figure 3 FIG. 1 is a schematic diagram of a terminal device provided by an embodiment of the present invention. Figure 3 As shown, a terminal device 3 of this embodiment includes: a processor 30, a memory 31, and a computer program 32 stored in the memory 31 and executable on the processor 30, such as a region recognition program for a graphene thermal therapy robot. When the processor 30 executes the computer program 32, the steps of the above-mentioned region recognition method of a graphene thermal therapy robot are implemented, such as Figure 1 Alternatively, when the processor 30 executes the computer program 32, the functions of each unit in the above-mentioned device embodiments are realized, for example, Figure 2 Function of the unit shown.

[0157] Exemplarily, the computer program 32 may be divided into one or more units, which are stored in the memory 31 and executed by the processor 30 to implement the present invention. The one or more units may be a series of computer program instruction segments capable of performing specific functions, which are used to describe the execution process of the computer program 32 in the terminal device 3. For example, the specific functions of each unit of the computer program 32 are as follows:

[0158] An acquisition unit, configured to acquire an original image on the thermal therapy platform and extract a human body region from the original image;

[0159] a segmentation unit, configured to segment the human body region based on a preset human body torso ratio to obtain images to be identified corresponding to a plurality of torso parts;

[0160] an input unit for extracting a target image to be identified corresponding to a target hyperthermia torso part, and inputting the target image to be identified into a torso probability recognition model to obtain a probability distribution map output by the torso probability recognition model; the probability distribution map is a probability data matrix composed of torso probabilities corresponding to a plurality of pixel points, the torso probability being the probability that a current pixel point is a pixel point corresponding to a target hyperthermia torso part; the target hyperthermia torso part being the torso part requiring hyperthermia treatment, and different hyperthermia torso parts corresponding to different torso probability recognition models;

[0161] An extraction unit, configured to extract a target image region corresponding to a target hyperthermia treatment torso part in the target image to be identified according to a probability value distribution in the probability distribution map;

[0162] The control unit is used to control the robot arm to move to the actual area corresponding to the target image area to perform the thermal treatment operation according to the conversion relationship between the image coordinate system and the actual coordinate system.

[0163] The terminal device includes but is not limited to a processor 30 and a memory 31. Those skilled in the art will understand that Figure 3 It is only an example of a terminal device 3 and does not constitute a limitation on a terminal device 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal device may also include input and output devices, network access devices, buses, etc.

[0164] The processor 30 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0165] The memory 31 may be an internal storage unit of the terminal device 3, such as a hard disk or memory of the terminal device 3. The memory 31 may also be an external storage device of the terminal device 3, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device 3. Furthermore, the memory 31 may include both an internal storage unit of the terminal device 3 and an external storage device. The memory 31 is used to store the computer program and other programs and data required by the roaming control device. The memory 31 may also be used to temporarily store data that has been output or is to be output.

[0166] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0167] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the embodiment of the method of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0168] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0169] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned various method embodiments can be implemented.

[0170] An embodiment of the present invention provides a computer program product. When the computer program product is run on a mobile terminal, the mobile terminal can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0171] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk.

[0172] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0173] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0174] In the embodiments provided by the present invention, it should be understood that the disclosed devices / network equipment and methods can be implemented in other ways. For example, the device / network equipment embodiments described above are merely illustrative. For example, the division of the modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0175] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units.

[0176] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0177] It will also be understood that the term "and / or" used in the present description and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0178] As used in the present specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to monitoring," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is monitored" may be interpreted as meaning "upon determination" or "in response to determining" or "upon monitoring [described condition or event]" or "in response to monitoring [described condition or event]," depending on the context.

[0179] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0180] References to "one embodiment" or "some embodiments" in the present specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present invention. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in yet other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0181] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A region recognition method for a graphene hyperthermia robot, characterized in that: The region recognition method of the graphene hyperthermia robot is applied to the graphene hyperthermia robot, which includes a robotic arm, a graphene hyperthermia module, a visual acquisition module, and a control module. The region recognition method of the graphene hyperthermia robot includes: collecting an original image on the thermal therapy platform and extracting a human body region in the original image; Based on a preset human torso ratio, the human body region is segmented to obtain images to be identified corresponding to a plurality of torso parts; Extracting a target image to be identified corresponding to a target hyperthermia torso part, and inputting the target image to be identified into a torso probability recognition model to obtain a probability distribution map output by the torso probability recognition model; the probability distribution map is a probability data matrix composed of torso probabilities corresponding to a plurality of pixel points, the torso probability being the probability that a current pixel point is a pixel point corresponding to a target hyperthermia torso part; the target hyperthermia torso part being the torso part requiring hyperthermia treatment, and different torso probability recognition models corresponding to different hyperthermia torso parts; extracting a target image region corresponding to a target hyperthermia treatment torso part in the target image to be identified according to the probability value distribution in the probability distribution graph; According to the conversion relationship between the image coordinate system and the actual coordinate system, the robotic arm is controlled to move to the actual area corresponding to the target image area to perform the thermal treatment operation.

2. The region recognition method of the graphene hyperthermia robot according to claim 1, characterized in that: The step of dividing the human body region into multiple images to be identified corresponding to the torso parts based on the preset human body torso proportions includes: Based on the first length direction of the human body region, dividing the human body region according to the preset human body torso ratio to obtain a plurality of first image regions; respectively acquiring first extended areas of the first image area in the length direction; splicing the first image area with a first extended area corresponding to the first image area to obtain a redundant image area; Inputting the redundant image area corresponding to the head-to-torso ratio into a head recognition model to obtain a recognition result output by the head recognition model; If the recognition result is the head area, the redundant image area is used as the image to be recognized; If the recognition result is a non-head region, segmenting the human body region according to the preset human body torso ratio based on the second length direction of the human body region to obtain a plurality of second image regions; respectively acquiring second extended areas of the second image area in the length direction; The second image area is spliced with a second extended area corresponding to the second image area to obtain the image to be recognized.

3. The region recognition method of the graphene hyperthermia robot according to claim 1, characterized in that: The steps of extracting the target image to be identified corresponding to the target hyperthermia treatment torso part, inputting the target image to be identified into the torso probability recognition model, and obtaining the probability distribution map output by the torso probability recognition model include: Scaling the target image to be identified to obtain target images to be identified at multiple scales; Inputting the target image to be identified into a first convolutional network to obtain first feature data output by the first convolutional network; the first feature data is used to characterize common features in the target image to be identified; The first feature data is input into a second convolutional network to obtain a probability distribution map output by the second convolutional network.

4. The region recognition method of the graphene hyperthermia robot according to claim 3, characterized in that: The step of inputting the first feature data into the second convolutional network to obtain a probability distribution map output by the second convolutional network includes: Inputting the first feature data into a plurality of sub-convolutional networks respectively to obtain a plurality of second feature data respectively output by the plurality of sub-convolutional networks; the plurality of sub-convolutional networks respectively use convolution kernels of different scales; Inputting the plurality of second feature data into a fully connected layer respectively to obtain a plurality of classification probabilities output by the fully connected layer; the fully connected layer is used to classify the target hyperthermia treatment torso part; Normalize multiple classification probabilities to obtain multiple target probabilities; Multiply each second feature data and the target probability corresponding to the second feature data to obtain third feature data; The plurality of third feature data are superimposed to obtain the probability distribution graph.

5. The region recognition method of the graphene hyperthermia robot according to claim 4, characterized in that: The step of superimposing a plurality of third feature data to obtain the probability distribution graph includes: Adding characteristic values of the same pixel position in the plurality of third characteristic data to obtain fourth characteristic data; extracting a maximum value from the fourth feature data; The probability distribution graph is obtained by dividing the characteristic value corresponding to each pixel position in the fourth characteristic data by the maximum value.

6. The region recognition method of the graphene hyperthermia robot according to claim 1, characterized in that: The step of extracting the target image area corresponding to the target hyperthermia treatment torso part in the target image to be identified according to the probability value distribution in the probability distribution graph comprises: Extract the probability distribution maps corresponding to multiple scales; Interpolating the probability distribution maps corresponding to the multiple scaling scales to obtain multiple interpolated probability distribution maps; the size of the interpolated probability distribution map is consistent with the original size of the target image to be recognized; the scaling scale refers to the size obtained by scaling the original size; Extracting a first pixel point whose probability value is greater than a first threshold value in the probability distribution map corresponding to the original size; Extracting a second pixel point having a probability value greater than a first threshold value from the interpolation probability distribution map; A target image area corresponding to a target hyperthermia treatment torso part in the target image to be identified is extracted according to the coordinate data of the first pixel point and the second pixel point.

7. The region recognition method of the graphene hyperthermia robot according to claim 6, characterized in that: The step of extracting the target image area corresponding to the target hyperthermia treatment torso part in the target image to be identified based on the coordinate data of the first pixel point and the second pixel point includes: Extracting image coordinate data corresponding to all first pixel points and all second pixel points; Counting the number of first pixel points or second pixel points corresponding to each of the image coordinate data; If the number of pixel points corresponding to the current image coordinate data is greater than a preset number, the pixel points corresponding to the current image coordinate data are used as pixel points corresponding to the target hyperthermia treatment torso part; The multiple pixel points corresponding to the target hyperthermia treatment torso area are used as the target image area.

8. A region recognition device for a graphene hyperthermia robot, characterized in that: The region recognition device of the graphene hyperthermia therapy robot includes: An acquisition unit, configured to acquire an original image on the thermal therapy platform and extract a human body region from the original image; a segmentation unit, configured to segment the human body region based on a preset human body torso ratio to obtain images to be identified corresponding to a plurality of torso parts; an input unit for extracting a target image to be identified corresponding to a target hyperthermia torso part, and inputting the target image to be identified into a torso probability recognition model to obtain a probability distribution map output by the torso probability recognition model; the probability distribution map is a probability data matrix composed of torso probabilities corresponding to a plurality of pixel points, the torso probability being the probability that a current pixel point is a pixel point corresponding to a target hyperthermia torso part; the target hyperthermia torso part being the torso part requiring hyperthermia treatment, and different hyperthermia torso parts corresponding to different torso probability recognition models; An extraction unit, configured to extract a target image region corresponding to a target hyperthermia treatment torso part in the target image to be identified according to a probability value distribution in the probability distribution map; The control unit is used to control the robot arm to move to the actual area corresponding to the target image area to perform the thermal treatment operation according to the conversion relationship between the image coordinate system and the actual coordinate system.

9. A terminal device, characterized in that: The terminal device includes: a memory, a processor, and a region recognition program of a graphene thermal therapy robot stored in the memory and executable on the processor. The region recognition program of the graphene thermal therapy robot is configured to implement the steps in the region recognition method of the graphene thermal therapy robot as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the region recognition method of the graphene thermal therapy robot are implemented.

Images