Power equipment dual-light intelligent vision temperature measurement method, system and equipment based on AI front-end technology, and storage medium

Abstract

The invention discloses a power equipment dual-light intelligent vision temperature measurement method, system and equipment based on an AI preposition technology, and a storage medium. The method comprises the steps of obtaining a color picture and an infrared image of a target range; carrying out space coordinate alignment on the color photo and the infrared image; performing histogram equalization processing on the color photo to obtain an equalized color photo; converting the temperature of each pixel point in the infrared image into a color space to obtain a color temperature image; fusing the balanced color photo and the color temperature image into an enhanced image; target detection is carried out on the enhanced image by adopting an AI model, and the AI model is used for identifying each target in the enhanced image; and for each identified target, carrying out temperature anomaly analysis and state identification according to the color of the target displayed in the enhanced image. According to the invention, a color photo and an infrared image are fused, the image quality is high, and the data processing efficiency is high.

Description

Technical Field

[0001] The present invention relates to intelligent visual temperature measurement technology, and in particular to a dual-light intelligent visual temperature measurement method, system, device and storage medium for power equipment based on AI front-end technology. Background Art

[0002] With the rapid development of new energy sites (such as wind power, photovoltaic power generation, etc.), the equipment in new energy sites is becoming increasingly complex, and the operating environment usually includes narrow monitoring space and limited field of view. Common image acquisition technologies include visible light imaging and infrared thermal imaging, but each has its own advantages and disadvantages: visible light imaging has advantages in detail performance, but it is difficult to penetrate obstructions; infrared thermal imaging can penetrate obstructions, but the image details are poor. In addition, traditional back-end processing solutions have high requirements for computing resources and slow processing speeds, which affects the real-time performance and accuracy of the system. Summary of the invention

[0003] In view of the problems existing in the prior art, the purpose of the present invention is to provide a dual-light intelligent visual temperature measurement method, system, equipment and storage medium for power equipment based on AI front-end technology, which combines dual-light imaging technology and AI front-end technology to improve image quality and data processing efficiency.

[0004] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0005] A dual-light intelligent visual temperature measurement method for power equipment based on AI front-end technology, comprising:

[0006] Acquire color photographs and infrared images of target ranges;

[0007] Align the spatial coordinates of color photos and infrared images;

[0008] Perform histogram equalization on color photos to obtain balanced color photos;

[0009] Convert the temperature of each pixel in the infrared image to the color space to obtain a color temperature image;

[0010] Fusion of the equalized color photo and the color temperature image into an enhanced image;

[0011] Using an AI model to perform target detection on the enhanced image, wherein the AI ​​model is used to identify each target in the enhanced image;

[0012] For each identified target, temperature anomaly analysis and status identification are performed according to the color of the target displayed in the enhanced image.

[0013] Furthermore, the step of performing histogram equalization on the color photo to obtain a balanced color photo specifically includes:

[0014] Perform histogram equalization on the color photo according to the following formula to obtain a balanced color photo:

[0015]

[0016] In the formula, I gs,vis (x, y) represents the pixel at coordinate (x, y) on the color photo before equalization. egs,vis (x,y) represents the pixel point with coordinates (x,y) on the balanced color photo, CDF() represents the cumulative distribution function, CDF min represents the minimum cumulative distribution value, L represents the number of gray levels, and H and W represent the horizontal and vertical lengths of the color photo.

[0017] Furthermore, the step of converting the temperature of each pixel in the infrared image into a color space to obtain a color temperature image specifically includes:

[0018] The temperature of each pixel in the infrared image is converted to the color space according to the following formula to obtain a color temperature image:

[0019]

[0020] In the formula, I mgs,ir (x, y) represents the pixel at coordinate (x, y) in the color temperature image, T gs,vir (x, y) represents the infrared image I before mapping gs,ir The temperature of the pixel at the coordinate (x, y), T max , T min They respectively represent the highest and lowest temperatures of all pixels in the infrared image before mapping, and Colormap() represents the color space mapping function.

[0021] Furthermore, the fusion of the balanced color photo and the color temperature image into an enhanced image specifically includes:

[0022] The image features of the color photo and the image features of the color temperature image will be extracted according to the following formula:

[0023] F egs,vis =SF(I egs,vis )

[0024] F mgs,ir =SF(I mgs,ir )

[0025] In the formula, F egs,vis Indicates balanced color photo I egs,vis The image features, F mgs,ir Represents the color temperature image I mgs,irImage features, SF() represents the image feature extraction algorithm;

[0026] The transformation matrix R is estimated as follows:

[0027]

[0028] In the formula, F egs,vis (i) F mgs,ir (i) respectively represent F egs,vis 、F mgs,ir The i-th feature point in ;

[0029] Image registration of color temperature images is performed according to the transformation matrix R:

[0030] I hmgs,ir =R·I mgs,ir

[0031] In the formula, I hmgs,ir represents the color temperature image after image registration;

[0032] Perform image fusion on the balanced color photo and the color temperature image after image registration to obtain an enhanced image:

[0033] I fused =α·I egs,vis +(1-α)I hmgs,ir

[0034] In the formula, I fused represents the enhanced image, and α is the weight parameter.

[0035] Furthermore, the AI ​​model includes:

[0036] An image input and preprocessing module, used to normalize the input enhanced image to a preset range;

[0037] A feature extraction module is used to extract high-dimensional features of the enhanced image using a convolutional neural network and generate a high-dimensional feature map;

[0038] A grid partitioning module is used to divide the high-dimensional feature map into several grids;

[0039] Bounding box prediction module, which is used to predict the bounding box and its confidence for each grid:

[0040] The category prediction module predicts the category to which the target belongs for each bounding box according to the confidence level, and the category is a pre-divided temperature range category.

[0041] Furthermore, for each identified target, temperature anomaly analysis and status recognition are performed according to the color of the target displayed in the enhanced image, specifically including:

[0042] For each object identified, look at the color it appears in the enhanced image to get the object's temperature and mark the areas where the object exceeds the temperature threshold:

[0043] A k ={(x,y)|T k (x,y)>T th}

[0044] Where, T th Indicates the temperature threshold, A k Indicates the area where target k exceeds the temperature threshold, T k (x,y) represents the temperature of the pixel at the target k coordinate (x,y);

[0045] Get the state of the target and output the state label S;

[0046] Integrate temperature and status information to generate early warning signals:

[0047] E=f(A k ,S)

[0048] In the formula, E represents the early warning signal, and f is the early warning judgment logic.

[0049] A dual-light intelligent visual temperature measurement system for power equipment based on AI front-end technology, including:

[0050] Synchronous contrast module, used to align the spatial coordinates of color photos and infrared images;

[0051] An equalization processing module is used to perform histogram equalization processing on the color photo to obtain a balanced color photo;

[0052] The temperature mapping module is used to convert the temperature of each pixel in the infrared image into a color space to obtain a color temperature image;

[0053] An image fusion module, used to fuse the balanced color photo and the color temperature image into an enhanced image;

[0054] An object detection module, used to perform object detection on the enhanced image using an AI model, wherein the AI ​​model is used to identify each object in the enhanced image;

[0055] The temperature analysis module is used to perform temperature anomaly analysis and status recognition for each identified target according to the color of the target displayed in the enhanced image.

[0056] Furthermore, the equalization processing module is specifically used to perform histogram equalization processing on the color photo according to the following formula to obtain a balanced color photo:

[0057]

[0058] In the formula, I gs,vis (x, y) represents the pixel at coordinate (x, y) on the color photo before equalization. egs,vis (x,y) represents the pixel point with coordinates (x,y) on the balanced color photo, CDF() represents the cumulative distribution function, CDF min represents the minimum cumulative distribution value, L represents the number of gray levels, and H and W represent the horizontal and vertical lengths of the color photo.

[0059] Furthermore, the temperature mapping module is specifically used to convert the temperature of each pixel in the infrared image into a color space according to the following formula to obtain a color temperature image:

[0060]

[0061] In the formula, I mgs,ir (x, y) represents the pixel at coordinate (x, y) in the color temperature image, T gs,vir (x, y) represents the infrared image I before mapping gs,ir The temperature of the pixel at the coordinate (x, y), T max 、T min They respectively represent the highest and lowest temperatures of all pixels in the infrared image before mapping, and Colormap() represents the color space mapping function.

[0062] Furthermore, the image fusion module specifically includes:

[0063] The feature extraction unit is used to extract the image features of the equalized color photo and the image features of the color temperature image according to the following formula:

[0064] F egs,vis =SF(I egs,vis )

[0065] F mgs,ir =SF(I mgs,ir )

[0066] In the formula, F egs,vis Indicates balanced color photo I egs,vis The image features, F mgs,ir Represents the color temperature image I mgs,ir Image features, SF() represents the image feature extraction algorithm;

[0067] The conversion matrix calculation unit is used to estimate the conversion matrix R according to the following formula:

[0068]

[0069] In the formula, F egs,vis (i) F mgs,ir(i) respectively represent F egs,vis 、F mgs,ir The i-th feature point in ;

[0070] The image registration unit is used to perform image registration on the color temperature image according to the transformation matrix R:

[0071] I hmgs,ir =R·I mgs,ir

[0072] In the formula, I hmgs,ir represents the color temperature image after image registration;

[0073] Perform image fusion on the balanced color photo and the color temperature image after image registration to obtain an enhanced image:

[0074] I fused =α·I egs,vis +(1-α)I hmgs,ir

[0075] In the formula, I fused represents the enhanced image, and α is the weight parameter.

[0076] Furthermore, the temperature analysis module specifically includes:

[0077] The anomaly detection unit is used to check the color displayed in the enhanced image for each identified target, thereby obtaining the temperature of the target and marking the area where the target exceeds the temperature threshold:

[0078] A k ={(x,y)|T k (x,y)>T th}

[0079] Where, T th Indicates the temperature threshold, A k Indicates the area where target k exceeds the temperature threshold, T k (x,y) represents the temperature of the pixel at the target k coordinate (x,y);

[0080] A state acquisition unit, used to acquire the state of the target and output a state label S;

[0081] The early warning unit is used to integrate temperature and status information to generate early warning signals:

[0082] E=f(A k ,S)

[0083] In the formula, E represents the early warning signal, and f is the early warning judgment logic.

[0084] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above method.

[0085] A computer-readable storage medium having a computer program / instruction stored thereon, wherein the computer program / instruction implements the above method when executed by a processor.

[0086] Compared with the prior art, the present invention has the following beneficial effects: before image fusion, the present invention processes the color photo and the infrared image separately, so that the fused enhanced image can better reflect the image features, increase the accuracy of the detection result, improve the quality of the visualized image, use the AI ​​model to identify the target in the enhanced image, and improve the data processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 It is a flow chart of a dual-light intelligent visual temperature measurement method for power equipment based on AI front-end technology provided by the present invention;

[0088] Figure 2 is a visualization diagram obtained by the method of the present invention;

[0089] Figure 3 This is an architecture diagram of the dual-light intelligent visual temperature measurement system for power equipment based on AI front-end technology provided by the present invention. DETAILED DESCRIPTION

[0090] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0091] Embodiment 1

[0092] The embodiment of the present invention provides a dual-light intelligent visual temperature measurement method for power equipment based on AI front-end technology, such as Figure 1 As shown, the following steps are included:

[0093] S101. Acquire color photos and infrared images of a target range.

[0094] Among them, a camera can be used to take a color photo, and the color photo taken by the camera is obtained, the color photo I vis is a two-dimensional color image matrix. An infrared thermal imager is used to capture the temperature distribution on the surface of the equipment and generate a thermal map, namely, an infrared image I ir , infrared image I ir is a single-channel grayscale image matrix.

[0095] S102, aligning the color photo and the infrared image in spatial coordinates.

[0096] After acquiring the color photo and infrared image, the timestamps are synchronized first, and then aligned in spatial coordinates to achieve image alignment to ensure time and space consistency. s,vis and infrared image I s, i r .

[0097] You can also use color photos later s,vis and infrared image I s,ir Filter and denoise according to the following formulas:

[0098] I gs,vis =G σ *I s,vis

[0099] I gs,ir =G σ *I s,ir

[0100] In the formula, I gs,vis represents the color photo after filtering and noise reduction, I gs,ir represents the infrared image after filtering and denoising, G σ represents a Gaussian kernel and * represents a convolution operation.

[0101] S103: Perform histogram equalization processing on the color photo to obtain a balanced color photo.

[0102] Specifically, the color photo can be processed by histogram equalization according to the following formula to obtain a balanced color photo:

[0103]

[0104] In the formula, I gs,vis (x, y) represents the pixel at the coordinate (x, y) on the color photo before equalization processing. The color photo before equalization processing is specifically the color photo after filtering and noise reduction. gs,vis , I egs,vis (x,y) represents the pixel point with coordinates (x,y) on the balanced color photo, CDF() represents the cumulative distribution function, CDF min represents the minimum cumulative distribution value, L represents the number of gray levels, and H and W represent the horizontal and vertical lengths of the color photo.

[0105] S104: Convert the temperature of each pixel in the infrared image into a color space to obtain a color temperature image.

[0106] Specifically, the temperature at each pixel in the infrared image can be converted to the color space according to the following formula to obtain a color temperature image:

[0107]

[0108] In the formula, I mgs,ir (x, y) represents the pixel at coordinate (x, y) in the color temperature image, T gs,vir (x, y) represents the temperature of the pixel at the coordinate (x, y) in the infrared image before mapping. The infrared image before mapping is the infrared image after filtering and denoising. gs,ir , T max 、T min They respectively represent the highest and lowest temperatures of all pixels in the infrared image before mapping, and Colormap() represents the color space mapping function.

[0109] S105: Fusing the balanced color photo and the color temperature image into an enhanced image.

[0110] This step specifically includes:

[0111] The image features of the color photo and the image features of the color temperature image will be extracted according to the following formula:

[0112] F egs,vis =SF(I egs,vis )

[0113] F mgs,ir =SF(I mgs,ir )

[0114] In the formula, F egs,vis Indicates balanced color photo I egs,vis The image features, F mgs,ir Represents the color temperature image I mgs,ir Image features, SF() represents the image feature extraction algorithm, such as SIFT (Scale-Invariant Feature Transform) algorithm;

[0115] The transformation matrix R is estimated as follows:

[0116]

[0117] In the formula, F egs,vis (i) F mgs,ir (i) respectively represent F egs,vis 、F mgs,ir The i-th feature point in ;

[0118] Image registration of color temperature images is performed according to the transformation matrix R:

[0119] I hmgs,ir =R·I mgs,ir

[0120] In the formula, I hmgs,irrepresents the color temperature image after image registration;

[0121] Perform image fusion on the balanced color photo and the color temperature image after image registration to obtain an enhanced image:

[0122] I fused =α·I egs,vis +(1-α)I hmgs,ir

[0123] In the formula, I fused represents the enhanced image, and α is the weight parameter.

[0124] S106. Use the AI ​​model to perform target detection on the enhanced image.

[0125] The AI ​​model is used to identify each target in the enhanced image. The AI ​​model includes:

[0126] An image input and preprocessing module, used to normalize the input enhanced image to a preset range;

[0127] A feature extraction module is used to extract high-dimensional features of the enhanced image using a convolutional neural network and generate a high-dimensional feature map;

[0128] A grid partitioning module is used to divide the high-dimensional feature map into several grids;

[0129] Bounding box prediction module, which is used to predict the bounding box and its confidence for each grid:

[0130] The category prediction module predicts the category to which the target belongs for each bounding box according to the confidence level, and the category is a pre-divided temperature range category.

[0131] Specifically, networks such as YOLOv5 can be used as AI models.

[0132] S107 . For each identified target, perform temperature anomaly analysis and state identification according to the color of the target displayed in the enhanced image.

[0133] This step specifically includes:

[0134] For each object identified, look at the color it appears in the enhanced image to get the object's temperature and mark the areas where the object exceeds the temperature threshold:

[0135] A k ={(x,y)|T k (x,y)>T th}

[0136] Where, T th Indicates the temperature threshold, A kIndicates the area where target k exceeds the temperature threshold, T k (x,y) represents the temperature of the pixel at the target k coordinate (x,y);

[0137] Get the state of the target and output the state label S; for example, if the target is a switch or a knife switch, the state label S∈{open, close} can be output.

[0138] Integrate temperature and status information to generate early warning signals:

[0139] E=f(A k ,S)

[0140] In the formula, E represents the early warning signal, and f is the early warning judgment logic.

[0141] Finally, after visualizing the recognition and detection results, Figure 2 shown.

[0142] Embodiment 2

[0143] Embodiment 2 of the present invention provides a dual-light intelligent visual temperature measurement system for power equipment based on AI front-end technology, which can be implemented in software and / or hardware, such as Figure 3 As shown, including:

[0144] An image acquisition module is used to obtain color photos and infrared images of the target range; the image acquisition module is specifically located in the data acquisition layer;

[0145] Synchronous contrast module, used to align the spatial coordinates of color photos and infrared images;

[0146] An equalization processing module is used to perform histogram equalization processing on the color photo to obtain a balanced color photo;

[0147] The temperature mapping module is used to convert the temperature of each pixel in the infrared image into a color space to obtain a color temperature image;

[0148] An image fusion module, used to fuse the balanced color photo and the color temperature image into an enhanced image;

[0149] The target detection module is used to perform target detection on the enhanced image using an AI model, wherein the AI ​​model is used to identify each target in the enhanced image; the synchronous contrast module, the equalization processing module, the temperature mapping module, the image fusion module, and the target detection module are all located in the AI ​​pre-processing layer;

[0150] The temperature analysis module is used to analyze the temperature anomaly and identify the status of each identified target according to the color displayed by the target in the enhanced image. The temperature analysis module is located in the intelligent analysis layer.

[0151] In addition, the system can also include an application service layer to perform real-time monitoring, fault warning, data statistics and remote diagnosis.

[0152] The system further includes a filtering and denoising module, which is used to filter and denoise the color photo and the infrared image according to the following formula:

[0153] I gs,vis =G σ *I s,vis

[0154] I gs,ir =G σ *I s,ir

[0155] In the formula, I s,vis ,I gs,vis Respectively represent the color photos before and after filtering and noise reduction, I s,ir ,I gs,ir Respectively represent the infrared images before and after filtering and denoising, G σ represents a Gaussian kernel and * represents a convolution operation.

[0156] The equalization processing module is specifically used to perform histogram equalization processing on the color photo according to the following formula to obtain a balanced color photo:

[0157]

[0158] In the formula, I gs,vis (x, y) represents the pixel at coordinate (x, y) on the color photo before equalization. egs,vis (x,y) represents the pixel point with coordinates (x,y) on the balanced color photo, CDF() represents the cumulative distribution function, CDF min represents the minimum cumulative distribution value, L represents the number of gray levels, and H and W represent the horizontal and vertical lengths of the color photo.

[0159] The temperature mapping module is specifically used to convert the temperature of each pixel in the infrared image into a color space according to the following formula to obtain a color temperature image:

[0160]

[0161] In the formula, I mgs,ir (x, y) represents the pixel at coordinate (x, y) in the color temperature image, T gs,vir (x,y) represents the infrared image i before mapping gs,ir The temperature of the pixel at the coordinate (x, y), T max , T minThey respectively represent the highest and lowest temperatures of all pixels in the infrared image before mapping, and Colormap() represents the color space mapping function.

[0162] Wherein, the image fusion module specifically includes:

[0163] The feature extraction unit is used to extract the image features of the equalized color photo and the image features of the color temperature image according to the following formula:

[0164] F egs,vis =SF(I egs,vis )

[0165] F mgs,ir =SF(I mgs,ir )

[0166] In the formula, F egs,vis Indicates balanced color photo I egs,vis The image features, F mgs,ir Represents the color temperature image I mgs,ir Image features, SF() represents the image feature extraction algorithm;

[0167] The conversion matrix calculation unit is used to estimate the conversion matrix R according to the following formula:

[0168]

[0169] In the formula, F egs,vis (i) F mgs,ir (i) respectively represent F egs,vis 、F mgs,ir The i-th feature point in ;

[0170] The image registration unit is used to perform image registration on the color temperature image according to the transformation matrix R:

[0171] I hmgs,ir =R·I mgs,ir

[0172] In the formula, I hmgs,ir represents the color temperature image after image registration;

[0173] Perform image fusion on the balanced color photo and the color temperature image after image registration to obtain an enhanced image:

[0174] I fused =α·I egs,vis +(1-α)I hmgs,ir

[0175] In the formula, I fused represents the enhanced image, and α is the weight parameter.

[0176] The AI ​​model includes:

[0177] An image input and preprocessing module, used to normalize the input enhanced image to a preset range;

[0178] A feature extraction module is used to extract high-dimensional features of the enhanced image using a convolutional neural network and generate a high-dimensional feature map;

[0179] A grid partitioning module is used to divide the high-dimensional feature map into several grids;

[0180] Bounding box prediction module, which is used to predict the bounding box and its confidence for each grid:

[0181] The category prediction module predicts the category to which the target belongs for each bounding box according to the confidence level, and the category is a pre-divided temperature range category.

[0182] Wherein, the temperature analysis module specifically includes:

[0183] The anomaly detection unit is used to check the color displayed in the enhanced image for each identified target, thereby obtaining the temperature of the target and marking the area where the target exceeds the temperature threshold:

[0184] A k ={(x,y)|T k (x,y)>T th}

[0185] Where, T th Indicates the temperature threshold, A k Indicates the area where target k exceeds the temperature threshold, T k (x,y) represents the temperature of the pixel at the target k coordinate (x,y);

[0186] A state acquisition unit, used to acquire the state of the target and output a state label S;

[0187] The early warning unit is used to integrate temperature and status information to generate early warning signals:

[0188] E=f(A k ,S)

[0189] In the formula, E represents the early warning signal, and f is the early warning judgment logic.

[0190] The system provided in the embodiment of the present invention can be used to execute the method provided in the first embodiment of the present invention, and has the corresponding functions and beneficial effects of the execution method.

[0191] It is worth noting that in the embodiment of the above system, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0192] The embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, or of course, it can be implemented only by hardware, as long as the function or effect can be achieved.

[0193] Embodiment 3

[0194] The embodiment of the present invention provides a computer device, which provides services for implementing the method of the first embodiment of the present invention. The device may include: a memory storing a computer executable program; a processor coupled to the memory; the processor calls the computer executable program stored in the memory to execute the steps in the method described in the first embodiment.

[0195] The memory may include computer system readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The device may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the memory may be used to read and write non-removable, non-volatile magnetic media (commonly referred to as "hard drives"). A program / utility having a set (at least one) of program modules may be stored in, for example, the memory, such program modules including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The computer executable program of the program module typically performs the functions and / or methods in the embodiments described in the present invention.

[0196] The processor executes various functional applications and data processing by running the program stored in the memory, such as implementing the method provided in the first embodiment of the present invention.

[0197] The code of the computer executable program can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" language or similar programming languages.

[0198] Embodiment 4

[0199] An embodiment of the present invention provides a storage medium including a computer executable program. When the computer executable program is executed by a computer processor, it is used to execute the method of the first embodiment.

[0200] The storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. Computer-readable media can be computer-readable signal media or computer-readable storage media. Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: electrical connections with one or more wires, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this document, computer-readable storage media can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0201] Of course, the storage medium including a computer executable program provided by an embodiment of the present invention, the computer executable program is not limited to the above method operations, and can also execute related operations in the method provided by any embodiment of the present invention.

[0202] It should be understood that the above embodiments and descriptions only describe the principles, main features and advantages of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, and these changes and improvements all fall within the scope of protection of the present invention.

Claims

1. A dual-light intelligent visual temperature measurement method for power equipment based on AI front-end technology, characterized in that: include: Acquire color photographs and infrared images of target ranges; Align the spatial coordinates of color photos and infrared images; Perform histogram equalization on color photos to obtain balanced color photos; Convert the temperature of each pixel in the infrared image to the color space to obtain a color temperature image; Fusion of the equalized color photo and the color temperature image into an enhanced image; Using an AI model to perform target detection on the enhanced image, wherein the AI ​​model is used to identify each target in the enhanced image; For each identified target, temperature anomaly analysis and status identification are performed according to the color of the target displayed in the enhanced image.

2. According to the AI ​​front-end technology-based dual-light intelligent visual temperature measurement method for power equipment in claim 1, it is characterized in that: The step of performing histogram equalization processing on the color photo to obtain a balanced color photo specifically includes: Perform histogram equalization on the color photo according to the following formula to obtain a balanced color photo: In the formula, I gs,vis (x, y) represents the pixel at coordinate (x, y) on the color photo before equalization. egs,vis (x,y) represents the pixel point with coordinates (x,y) on the balanced color photo, CDF() represents the cumulative distribution function, CDF min represents the minimum cumulative distribution value, L represents the number of gray levels, and H and W represent the horizontal and vertical lengths of the color photo.

3. According to the AI ​​front-end technology-based dual-light intelligent visual temperature measurement method for power equipment in claim 1, it is characterized in that: The step of converting the temperature of each pixel in the infrared image into a color space to obtain a color temperature image specifically includes: The temperature of each pixel in the infrared image is converted to the color space according to the following formula to obtain a color temperature image: In the formula, I mgs,ir (x, y) represents the pixel at coordinate (x, y) in the color temperature image, T gs,vir (x,y) represents the infrared image i before mapping gs,ir The temperature of the pixel at the coordinate (x, y), T max , T min They respectively represent the highest and lowest temperatures of all pixels in the infrared image before mapping, and Colormap() represents the color space mapping function.

4. According to the AI ​​front-end technology-based dual-light intelligent visual temperature measurement method for power equipment in claim 1, it is characterized in that: The step of fusing the balanced color photo and the color temperature image into an enhanced image specifically includes: The image features of the color photo and the image features of the color temperature image will be extracted according to the following formula: F egs,vis =SF(I egs,vis ) F mgs,ir =SF(I mgs,ir ) In the formula, F egs,vis Indicates balanced color photo I egs,vis The image features, F mgs,ir Represents the color temperature image I mgs,ir Image features, SF() represents the image feature extraction algorithm; The transformation matrix R is estimated as follows: In the formula, F egs,vis (i) F mgs,ir (i) respectively represent F egs,vis 、F mgs,ir The i-th feature point in ; Image registration of color temperature images is performed according to the transformation matrix R: I hmgs,ir =R·I mgs,ir In the formula, I hmgs,ir represents the color temperature image after image registration; Perform image fusion on the balanced color photo and the color temperature image after image registration to obtain an enhanced image: I fused =α·I egs,vis +(1-α)I hmgs,ir In the formula, I fused represents the enhanced image, and α is the weight parameter.

5. According to the AI ​​front-end technology-based dual-light intelligent visual temperature measurement method for power equipment in claim 1, it is characterized in that: The AI ​​model includes: An image input and preprocessing module, used to normalize the input enhanced image to a preset range; A feature extraction module, for extracting high-dimensional features of the enhanced image using a convolutional neural network and generating a high-dimensional feature map; A grid partitioning module is used to divide the high-dimensional feature map into several grids; Bounding box prediction module, which is used to predict the bounding box and its confidence for each grid: The category prediction module predicts the category to which the target belongs for each bounding box according to the confidence level, and the category is a pre-divided temperature range category.

6. According to the AI ​​front-end technology-based dual-light intelligent visual temperature measurement method for power equipment in claim 1, it is characterized in that: For each identified target, temperature anomaly analysis and status recognition are performed according to the color of the target displayed in the enhanced image, specifically including: For each object identified, look at the color it appears in the enhanced image to get the object's temperature and mark the areas where the object exceeds the temperature threshold: A k ={(x,y)|T k (x,y)>T th } Where, T th Indicates the temperature threshold, A k Indicates the area where target k exceeds the temperature threshold, T k (x,y) represents the temperature of the pixel at the target k coordinate (x,y); Get the state of the target and output the state label S; Integrate temperature and status information to generate early warning signals: E=f(A k ,S) In the formula, E represents the early warning signal, and f is the early warning judgment logic.

7. A dual-light intelligent visual temperature measurement system for power equipment based on AI front-end technology, characterized in that: include: An image acquisition module, used to obtain color photos and infrared images of the target range; Synchronous contrast module, used to align the spatial coordinates of color photos and infrared images; An equalization processing module is used to perform histogram equalization processing on the color photo to obtain a balanced color photo; The temperature mapping module is used to convert the temperature of each pixel in the infrared image into a color space to obtain a color temperature image; An image fusion module, used to fuse the balanced color photo and the color temperature image into an enhanced image; An object detection module, used to perform object detection on the enhanced image using an AI model, wherein the AI ​​model is used to identify each object in the enhanced image; The temperature analysis module is used to perform temperature anomaly analysis and status recognition for each identified target according to the color of the target displayed in the enhanced image.

8. The dual-light intelligent visual temperature measurement system for power equipment based on AI front-end technology according to claim 7 is characterized in that: The equalization processing module is specifically used to perform histogram equalization processing on the color photo according to the following formula to obtain a balanced color photo: In the formula, I gs,vis (x, y) represents the pixel at coordinate (x, y) on the color photo before equalization. egs,vis (x,y) represents the pixel point with coordinates (x,y) on the balanced color photo, CDF() represents the cumulative distribution function, CDF min represents the minimum cumulative distribution value, L represents the number of gray levels, and H and W represent the horizontal and vertical lengths of the color photo.

9. The dual-light intelligent visual temperature measurement system for power equipment based on AI front-end technology according to claim 7 is characterized in that: The temperature mapping module is specifically used to convert the temperature of each pixel in the infrared image into a color space according to the following formula to obtain a color temperature image: In the formula, I mgs,ir (x, y) represents the pixel at coordinate (x, y) in the color temperature image, T gs,vir (x, y) represents the infrared image I before mapping gs,ir The temperature of the pixel at the coordinate (x, y), T max , T min They respectively represent the highest and lowest temperatures of all pixels in the infrared image before mapping, and Colormap() represents the color space mapping function.

10. The dual-light intelligent visual temperature measurement system for power equipment based on AI front-end technology according to claim 7 is characterized in that: The image fusion module specifically includes: The feature extraction unit is used to extract the image features of the equalized color photo and the image features of the color temperature image according to the following formula: F egs,vis =SF(I egs,vis ) F mgs,ir =SF(I mgs,ir ) In the formula, F egs,vis Indicates balanced color photo I egs,vis The image features, F mgs,ir Represents the color temperature image I mgs,ir Image features, SF() represents the image feature extraction algorithm; The conversion matrix calculation unit is used to estimate the conversion matrix R according to the following formula: In the formula, F egs,vis (i) F mgs,ir (i) respectively represent F egs,vis 、F mgs,ir The i-th feature point in ; The image registration unit is used to perform image registration on the color temperature image according to the transformation matrix R: I hmgs,ir =R·I mgs,ir In the formula, I hmgs,ir represents the color temperature image after image registration; Perform image fusion on the balanced color photo and the color temperature image after image registration to obtain an enhanced image: I fused =α·I egs,vis +(1-α)I hmgs,ir In the formula, I fused represents the enhanced image, and α is the weight parameter.

11. The dual-light intelligent visual temperature measurement system for power equipment based on AI front-end technology according to claim 7 is characterized in that: The AI ​​model includes: An image input and preprocessing module, used to normalize the input enhanced image to a preset range; A feature extraction module, for extracting high-dimensional features of the enhanced image using a convolutional neural network and generating a high-dimensional feature map; A grid partitioning module is used to divide the high-dimensional feature map into several grids; Bounding box prediction module, which is used to predict the bounding box and its confidence for each grid: The category prediction module predicts the category to which the target belongs for each bounding box according to the confidence level, and the category is a pre-divided temperature range category.

12. The dual-light intelligent visual temperature measurement system for power equipment based on AI front-end technology according to claim 7 is characterized in that: The temperature analysis module specifically includes: The anomaly detection unit is used to check the color displayed in the enhanced image for each identified target, thereby obtaining the temperature of the target and marking the area where the target exceeds the temperature threshold: A k ={(x,y)|T k (x,y)>T th } Where, T th Indicates the temperature threshold, A k Indicates the area where target k exceeds the temperature threshold, T k (x,y) represents the temperature of the pixel at the target k coordinate (x,y); A state acquisition unit, used to acquire the state of the target and output a state label S; The early warning unit is used to integrate temperature and status information to generate early warning signals: E=f(A k ,S) In the formula, E represents the early warning signal, and f is the early warning judgment logic.

13. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: The processor executes the computer program to implement the method according to any one of claims 1 to 6.

14. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: The computer program / instructions, when executed by a processor, implement the method of any one of claims 1-6.

Images