Infrared temperature measurement and recording methods and systems

Abstract

This invention provides an infrared temperature measurement and recording method and system, relating to the field of equipment inspection technology. The method first acquires a first image and thermal imaging data; then, it preprocesses the first image and extracts multiple edge features from the preprocessed image; next, based on the multiple edge features, it generates an index vector using a retrieval model, and retrieves candidate devices from an equipment database based on the index vector; finally, in response to a confirmation command, it stores the thermal imaging data in the monitoring report of the candidate devices. Because this invention employs edge feature extraction, the amount of image data processing is greatly reduced, resulting in strong identification. The use of a retrieval model to generate index vectors facilitates rapid searching of the equipment database. Furthermore, modern methods can be used to automatically perform the above operations, eliminating the need for manual recording of thermal imaging data and improving inspection efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of equipment inspection, and in particular to an infrared temperature measurement recording method and system. BACKGROUND

[0002] Adopting infrared temperature measurement for substation equipment inspection can greatly improve the inspection efficiency and quickly find some hard-to-detect equipment problems.

[0003] In the prior art, after monitoring the measurement point by using an infrared temperature measuring instrument, the name, model and number of the detection point need to be checked, and the equipment inspection is recorded according to the name, model, number and measurement result of the measurement point, and a monitoring report is formed.

[0004] This way, due to manual checking and filling, after measuring each measurement point, the measurement result needs to be recorded first, and then the measurement of the next point can be performed, so the measurement and recording efficiency is low.

[0005] Therefore, it is necessary to develop an infrared temperature measurement recording method to solve the problems in the prior art. SUMMARY

[0006] The embodiments of the present application provide an infrared temperature measurement recording method and system to solve the problem of low efficiency of manual point-by-point recording of infrared temperature measurement in the prior art.

[0007] In a first aspect, the embodiments of the present application provide an infrared temperature measurement recording method, comprising:

[0008] obtaining a first image and thermal imaging data, wherein the first image and the thermal imaging data are generated based on the same measurement point;

[0009] preprocessing the first image, and extracting a plurality of edge features from the preprocessed first image;

[0010] generating an index vector by using a retrieval model according to the plurality of edge features, and obtaining a candidate device from a device archive according to the index vector;

[0011] in response to a confirmation instruction, storing the thermal imaging data in a monitoring report of the candidate device.

[0012] In a possible implementation manner, the preprocessing of the first image comprises:

[0013] desaturating the first image to obtain a gray image;

[0014] counting distribution intervals of a plurality of pixel gray values in the gray image;

[0015] According to the distribution interval and the distribution quantity of the pixel gray values, the pixel gray values are adjusted so as to make the distribution interval and the distribution quantity of the pixel gray values respectively conform to a preset interval and a preset distribution quantity.

[0016] In a possible implementation, the adjusting the pixel gray values according to the distribution interval and the distribution quantity of the pixel gray values so as to make the distribution interval and the distribution quantity of the pixel gray values respectively conform to a preset interval and a preset distribution quantity comprises:

[0017] a distribution interval adjusting step of adjusting the pixel gray values according to a first formula, the preset interval and the distribution interval of the pixel gray values, wherein the first formula is:

[0018]

[0019] wherein, pixel (rn) is the gray value after the rth iteration adjustment, pixel max is the maximum value in the pixel gray values, pixel min is the minimum value in the pixel gray values, v max is the upper limit value of the preset interval, v min is the lower limit value of the preset interval.

[0020] a step of counting the quantity of the pixel gray values distributed in a plurality of sub-intervals after the adjustment, to obtain a plurality of distribution quantities;

[0021] if the deviation of the plurality of distribution quantities from the preset distribution quantity is greater than a threshold value, determining a target sub-interval according to the plurality of distribution quantities and the preset distribution quantity, adjusting the pixel gray values according to a second formula and the target sub-interval, and jumping to the distribution interval adjusting step, wherein the target sub-interval is the interval with the maximum deviation of the distribution quantity from the preset distribution quantity, and the second formula is:

[0022] pixel (rn+1) = w p · DIF · (pixel (rn) - ITV) 2 + C

[0023] wherein, pixel (rn+1) is the pixel gray value after the (r+1)th iteration adjustment, w p is an adjustment coefficient, DIF is the difference between the distribution quantity of the target sub-interval and the preset distribution quantity, ITV is the median value of the target sub-interval, and C is a bias constant.

[0024] In a possible implementation, the plurality of edge features are extracted from the preprocessed first image, including:

[0025] An edge extraction operator is acquired.

[0026] A first edge feature map is obtained from the edge extraction operator and the preprocessed first image by means of convolution calculation.

[0027] The first edge feature map is edge cropped and segmented to obtain a plurality of second edge feature maps.

[0028] The plurality of second edge feature maps are subjected to a pooling operation according to a preset size to obtain a plurality of third feature maps.

[0029] The plurality of third feature maps are taken as the plurality of edge features.

[0030] In a possible implementation, the first edge feature map is obtained from the edge extraction operator and the preprocessed first image by means of convolution calculation, including:

[0031] An index indication, a horizontal edge operator matrix, and a vertical edge operator matrix are acquired, where the horizontal edge operator matrix and the vertical edge operator matrix are of the same type.

[0032] A matrix taking step is performed, in which a first matrix is taken from the preprocessed first image according to the index indication, where the first matrix is of the same type as the horizontal edge operator matrix and the vertical edge operator matrix.

[0033] Horizontal edge features and vertical edge features are extracted respectively according to a third formula, the first matrix, the horizontal edge operator matrix, and the vertical edge operator matrix, where the third formula is:

[0034]

[0035] where EF is the horizontal edge feature or the vertical edge feature, PM mn is an element in the mth row and the nth column of the first matrix, FM mn is an element in the mth row and the nth column of the horizontal edge operator matrix or the vertical edge operator matrix, M is the total number of rows of the horizontal edge operator matrix or the vertical edge operator matrix, and N is the total number of columns of the horizontal edge operator matrix or the vertical edge operator matrix.

[0036] The horizontal edge features and the vertical edge features are added to a horizontal feature matrix and a vertical feature matrix respectively according to the index indication.

[0037] If the index indicates that the end of the pre-processed first image has not been reached, the index is offset according to a preset order, and the matrix obtaining step is jumped to;

[0038] Otherwise, a sum matrix of the horizontal feature matrix and the vertical feature matrix is taken as a sum feature matrix;

[0039] Values in the sum feature matrix that are lower than an edge threshold are set to zero, to obtain the first edge feature map.

[0040] In a possible implementation, the edge cropping and segmentation of the first edge feature map to obtain a plurality of second edge feature maps comprises:

[0041] A plurality of rows and columns in the first edge feature map in which the number of consecutive zero values exceeds a consecutive threshold are obtained as a plurality of segmentation rows and a plurality of segmentation columns;

[0042] The first edge feature map is segmented according to the segmentation rows and the plurality of segmentation columns, to obtain a plurality of intermediate feature maps;

[0043] Intermediate feature maps in the plurality of intermediate feature maps that are smaller than the preset size are deleted, and the remaining plurality of intermediate feature maps are taken as the plurality of second edge feature maps.

[0044] In a possible implementation, the generating an index vector using a retrieval model according to the plurality of edge features comprises:

[0045] An edge feature library is obtained, the edge feature library comprising a plurality of edge feature samples, and each edge feature sample being provided with an identifier;

[0046] The plurality of edge features are respectively input into the retrieval model, and a plurality of target identifiers are obtained by searching the edge feature library using the retrieval model, wherein each target identifier is an identifier of an edge feature sample that is closest to the plurality of edge features;

[0047] The plurality of target identifiers are arranged according to the values, and an index vector is constructed according to the arranged plurality of target identifiers.

[0048] In a possible implementation, the edge feature and the edge feature sample are both matrices, and the retrieval model is:

[0049]

[0050] In the formula, PS is a similarity coefficient, SM ab is an element in the a th row and the b th column of the edge feature matrix, and BM abis an element in the a-th row and the b-th column of the edge feature sample matrix, A is the total number of rows of the edge feature matrix, and B is the total number of columns of the edge feature matrix.

[0051] In a possible implementation, the inputting of the plurality of edge features into the retrieval model respectively, the retrieval in the edge feature library through the retrieval model, and the obtaining of the plurality of target identifications include the following steps.

[0052] For each edge feature in the plurality of edge features, the following steps are performed.

[0053] The edge feature and each edge feature sample in the plurality of edge feature samples are input into the retrieval model to obtain a plurality of similarity coefficients, wherein the plurality of similarity coefficients correspond to the plurality of edge feature samples.

[0054] The maximum similarity coefficient in the plurality of similarity coefficients is selected as a candidate coefficient.

[0055] If the candidate coefficient is greater than a similarity threshold, an identification of an edge feature sample corresponding to the candidate coefficient is taken as a target identification.

[0056] In a second aspect, an embodiment of the present application provides an infrared temperature measurement recording system, comprising:

[0057] An infrared temperature measurement device and a server, wherein the infrared temperature measurement device is in communication connection with the server.

[0058] The infrared temperature measurement device collects a first image and thermal imaging data of a measurement point, and sends the first image and the thermal imaging data to the server.

[0059] After the server receives the first image and the thermal imaging data, the server performs the steps of the infrared temperature measurement recording method according to the first aspect or any possible implementation of the first aspect.

[0060] Compared with the prior art, the embodiment of the present application has the following beneficial effects:

[0061] The embodiment of the present application provides an infrared temperature measurement recording method and system, first, a first image and thermal imaging data are acquired, wherein the first image and the thermal imaging data are generated based on the same measurement point; then, the first image is preprocessed, and a plurality of edge features are extracted from the preprocessed first image; then, according to the plurality of edge features, an index vector is generated by using a retrieval model, and a candidate device is acquired from a device archive according to the index vector; finally, in response to a confirmation instruction, the thermal imaging data is stored in a monitoring report of the candidate device. The embodiment of the present application is based on the extraction of image edges, generates a plurality of features, and then uses a retrieval model to generate an index vector from the plurality of features, finds a candidate device according to the index vector, and finally stores the thermal imaging data in the monitoring report of the candidate device. Since the edge feature extraction operation is adopted, the image data processing amount is greatly reduced, the identification is strong, the retrieval model is used to generate the index vector, the device archive can be quickly searched, and some modern means, for example, computer technology, can automatically perform the above operation, without manual recording of the thermal imaging data, and the inspection efficiency is improved. BRIEF DESCRIPTION OF DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0063] Figure 1 It is the implementation flowchart of the infrared temperature measurement recording method provided by the embodiment of the present application;

[0064] Figure 2 It is the principle diagram of the edge cutting and segmentation process of the edge feature map provided by the embodiment of the present application;

[0065] Figure 3 It is the structure schematic diagram of the infrared temperature measurement recording system provided by the embodiment of the present application. DETAILED DESCRIPTION

[0066] In the following description, specific details such as specific system structures, techniques, etc. are presented in order to thoroughly understand the embodiments of the present application. However, it should be clear to those skilled in the art that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits and methods are omitted to avoid unnecessary details that hinder the description of the present application.

[0067] In order to make the objects, technical solutions and advantages of the present application clearer, the following will be described by specific embodiments in combination with the drawings.

[0068] Figure 1 The implementation flowchart of the infrared temperature measurement recording method provided by the embodiment of the present application is described in detail as follows:

[0069] In step 101, a first image and thermal imaging data are acquired, wherein the first image and the thermal imaging data are generated based on the same measurement point.

[0070] In step 102, the first image is preprocessed, and a plurality of edge features are extracted from the preprocessed first image.

[0071] In some embodiments, step 102 includes:

[0072] The first image is desaturated to obtain a grayscale image.

[0073] The distribution intervals of a plurality of pixel grayscale values in the grayscale image are counted.

[0074] The plurality of pixel grayscale values are adjusted according to the distribution intervals and the distribution quantities of the plurality of pixel grayscale values, so that the distribution intervals and the distribution quantities of the plurality of pixel grayscale values respectively meet preset intervals and preset distribution quantities.

[0075] In some embodiments, the adjusting the plurality of pixel grayscale values according to the distribution intervals and the distribution quantities of the plurality of pixel grayscale values, so that the distribution intervals and the distribution quantities of the plurality of pixel grayscale values respectively meet preset intervals and preset distribution quantities, includes:

[0076] A distribution interval adjustment step: adjusting the plurality of pixel grayscale values according to a first formula, the preset intervals, and the distribution intervals of the plurality of pixel grayscale values, wherein the first formula is:

[0077]

[0078] In the formula, pixel (rn) is the grayscale value after the nth iteration adjustment, pixel max is the maximum value in the plurality of pixel grayscale values, pixel min is the minimum value in the plurality of pixel grayscale values, v max is the upper limit value of the preset interval, v min is the lower limit value of the preset interval.

[0079] The quantities of the plurality of pixel grayscale values distributed in a plurality of subintervals are counted to obtain a plurality of distribution quantities.

[0080] If the deviation of the plurality of distribution quantities from the preset distribution quantity is greater than a threshold value, a target sub-interval is determined according to the plurality of distribution quantities and the preset distribution quantity, the plurality of pixel gray values are adjusted according to a second formula and the target sub-interval, and the distribution interval adjustment step is jumped to, wherein the target sub-interval is an interval with the largest deviation of the distribution quantity from the preset distribution quantity, and the second formula is:

[0081] pixel (rn+1) = w p · DIF · (pixel (rn) - ITV) 2 + C

[0082] In the formula, pixel (rn+1) is the pixel gray value after the rn+1th iteration adjustment, w p is an adjustment coefficient, DIF is the difference between the target sub-interval distribution quantity and the preset distribution quantity, ITV is the median of the target sub-interval, and C is a bias constant.

[0083] In some embodiments, a plurality of edge features are extracted from the preprocessed first image, including:

[0084] An edge extraction operator is obtained.

[0085] A first edge feature map is obtained by convolution calculation according to the edge extraction operator and the preprocessed first image.

[0086] The first edge feature map is edge-cutting and segmented to obtain a plurality of second edge feature maps.

[0087] The plurality of second edge feature maps are subjected to a pooling operation according to a preset size to obtain a plurality of third feature maps.

[0088] The plurality of third feature maps are taken as the plurality of edge features.

[0089] In some embodiments, the first edge feature map is obtained by convolution calculation according to the edge extraction operator and the preprocessed first image, including:

[0090] An index indication, a horizontal edge operator matrix and a vertical edge operator matrix are obtained, wherein the horizontal edge operator matrix and the vertical edge operator matrix are of the same type.

[0091] A matrix taking step: a first matrix is taken out from the preprocessed first image according to the index indication, wherein the first matrix is of the same type as the horizontal edge operator matrix and the vertical edge operator matrix.

[0092] According to a third formula, the first matrix, the horizontal edge operator matrix and the vertical edge operator matrix, horizontal edge features and vertical edge features are extracted respectively, wherein the third formula is:

[0093]

[0094] In the formula, EF is the horizontal edge feature or the vertical edge feature, PM mn is an element in the mth row and the nth column of the first matrix, FM mn is an element in the mth row and the nth column of the horizontal edge operator matrix or the vertical edge operator matrix, M is the total number of rows of the horizontal edge operator matrix or the vertical edge operator matrix, and N is the total number of columns of the horizontal edge operator matrix or the vertical edge operator matrix.

[0095] According to the index indication, the horizontal edge features and the vertical edge features are added into a horizontal feature matrix and a vertical feature matrix respectively.

[0096] If the index indication does not reach the end of the preprocessed first image, the index indication is offset according to a preset order, and the matrix extraction step is jumped to.

[0097] Otherwise, a sum matrix of the horizontal feature matrix and the vertical feature matrix is taken as a sum feature matrix.

[0098] Values lower than an edge threshold in the sum feature matrix are set to zero, and the first edge feature map is obtained.

[0099] In some embodiments, the first edge feature map is edge cropped and segmented to obtain a plurality of second edge feature maps, including:

[0100] A plurality of rows and a plurality of columns with a continuous zero value number exceeding a continuous threshold are obtained from the first edge feature map as a plurality of segmentation rows and a plurality of segmentation columns.

[0101] According to the segmentation rows and the plurality of segmentation columns, the first edge feature map is segmented to obtain a plurality of intermediate feature maps.

[0102] Intermediate feature maps smaller than the preset size are deleted from the plurality of intermediate feature maps, and the remaining plurality of intermediate feature maps are taken as the plurality of second edge feature maps.

[0103] Exemplarily, in the embodiment of the present application, the infrared temperature measuring device is used to detect the measuring point, and an image and a thermal image of the measuring point are obtained at the same time. In order to record the data of the thermal image into the inspection record, the image is identified based on the obtained image in the embodiment of the present application, so that the identification of the detection point is determined, and the thermal image data is stored into the corresponding inspection record according to the identification of the monitoring point, thereby improving the efficiency of the inspection.

[0104] In order to achieve the above effect, the image obtained is preprocessed in the embodiment of the present application, and the edge features of the image are extracted after the preprocessing.

[0105] In the aspect of image preprocessing, the image is first desaturated to form a gray image. Due to the influence of factors such as light intensity, shooting angle, etc., the numerical distribution of the gray image will have a large difference. In order to more accurately extract the edge features subsequently, the distribution interval of the gray value of the pixel in the gray image and the number of the distribution of the pixel gray value are iterated repeatedly in the embodiment of the present application, so that the distribution interval and the number of the gray value of the gray image are the same as the expected distribution.

[0106] In the iteration, the gray value is first adjusted to a preset interval by using a first formula:

[0107]

[0108] In the formula, pixel (rn) is the gray value adjusted in the nth iteration, pixel max is the maximum value of the gray value of the pixel, pixel min is the minimum value of the gray value of the pixel, v max is the upper limit value of the preset interval, v min is the lower limit value of the preset interval.

[0109] After the pixel value is adjusted, the interval is divided into a plurality of subintervals. For example, the pixel value is distributed between 10-210, which is divided into four subintervals of 10-60, 60-100, 100-160 and 160-210. The purpose of the different spans of the four subintervals is to make the number of pixels in each subinterval the same. For example, if a certain gray image has a total of 1 million pixels, the reasonable distribution should be 250,000 pixels in each interval. If the deviation is large, for example, the four subintervals are distributed as 200,000, 230,000, 350,000 and 220,000, it is obvious that the number of pixels in the third subinterval deviates too much and should be adjusted. The second formula is used to adjust in the embodiment of the present application:

[0110] pixel (rn+1) = w p · DIF · (pixel (rn) - ITV)2 +C

[0111] In the formula, pixel (rn+1) is the pixel gray value after the nth+1 iteration adjustment, w p is an adjustment coefficient, DIF is the difference between the target sub-interval distribution quantity and the preset distribution quantity, ITV is the median of the target sub-interval, and C is a bias constant.

[0112] After adjustment, the overall interval distribution will be offset, at which time it is necessary to return to the first formula to readjust the overall interval. This is repeated until the distribution of the overall interval and the pixel quantity is less than a threshold value.

[0113] For edge feature extraction, first, the above-processed gray scale image is used to extract edges using an edge extraction operator. The horizontal edge operator is used in the embodiments of the present application:

[0114]

[0115] and the vertical edge operator:

[0116]

[0117] The horizontal edge feature and the vertical edge feature are extracted by convolution, and the overall edge is extracted after the two features are fused. Finally, the overall edge is cut and pooled to obtain multiple edge features.

[0118] For convolution operation, in one application scenario, a data block is sequentially taken out from left to right and from top to bottom from the gray scale image to construct a matrix, and the horizontal edge feature and the vertical edge feature contained in the matrix are calculated using the third formula, the above-mentioned horizontal edge operator and the vertical edge operator:

[0119]

[0120] In the formula, EF is the horizontal edge feature or the vertical edge feature, PM mn is the element in the mth row and the nth column of the first matrix, FM mn is the element in the mth row and the nth column of the horizontal edge operator matrix or the vertical edge operator matrix, M is the total number of rows of the horizontal edge operator matrix or the vertical edge operator matrix, and N is the total number of columns of the horizontal edge operator matrix or the vertical edge operator matrix.

[0121] The obtained horizontal edge feature and vertical edge feature are respectively stored in the horizontal feature matrix and the vertical feature matrix according to the position of the data block taken out.

[0122] The horizontal feature matrix and the vertical feature matrix are added to obtain a sum matrix, and then the values less than an edge threshold are set to zero by using the edge threshold to obtain a fused edge matrix.

[0123] Since the edge matrix can include edges of multiple objects or edges of multiple parts of an object, for example, an instrument includes a display part and a key area, the edge matrix obtained by the above steps is multiple independent and closed edges, and at this time, the multiple independent and closed edges need to be segmented to form multiple independent features.

[0124] As shown in FIG. 2, Figure 2 which shows a situation after edges are extracted. The edge matrix 201 includes multiple independent features 204, which are segmented by blank lines 202. The blank lines are actually rows or columns of continuous zeros. In the embodiments of the present application, the length of the blank lines is required to be not less than a certain threshold, for example, the short line 203 is not recognized as a blank line due to its too short length.

[0125] The features extracted by the lines can be regarded as matrix blocks, and the sizes of the matrix blocks are different. In order to unify the sizes of the matrix blocks, a pooling operation is performed in the embodiments of the present application. In one application scenario, the target after pooling is 600x800, and the size of the matrix block is 700x1000. The maximum pooling is used, and at this time, a 100x200 data block is taken out from the matrix block each time, the maximum value is selected for pooling, and the result after pooling is the size of the target.

[0126] In step 103, an index vector is generated by using a retrieval model according to the plurality of edge features, and a candidate device is obtained from a device library according to the index vector.

[0127] In some embodiments, step 103 includes:

[0128] An edge feature library is obtained, and the edge feature library includes a plurality of edge feature samples, and each edge feature sample is provided with an identifier.

[0129] The plurality of edge features are input into the retrieval model respectively, and a plurality of target identifiers are obtained by searching the edge feature library by using the retrieval model, wherein the target identifier is the identifier of the edge feature sample closest to the plurality of edge features.

[0130] The plurality of target identifiers are arranged according to the size of the values, and an index vector is constructed according to the plurality of target identifiers after arrangement.

[0131] In some embodiments, the edge feature and the edge feature sample are both matrices, and the retrieval model is:

[0132]

[0133] wherein, PS is a similarity coefficient, SM ab is an element in the a-th row and b-th column of the edge feature matrix, Bm ab is an element in the a-th row and b-th column of the edge feature sample matrix, A is the total number of rows of the edge feature matrix, and B is the total number of columns of the edge feature matrix.

[0134] In some embodiments, the inputting of the plurality of edge features into the retrieval model respectively, the retrieval in the edge feature library through the retrieval model, and the obtaining of the plurality of target identifications, comprises:

[0135] For each edge feature in the plurality of edge features, the following steps are performed:

[0136] The edge feature and each edge feature sample in the plurality of edge feature samples are inputted into the retrieval model to obtain a plurality of similarity coefficients, wherein the plurality of similarity coefficients correspond to the plurality of edge feature samples;

[0137] The maximum value of the plurality of similarity coefficients is selected as a candidate coefficient;

[0138] If the candidate coefficient is greater than a similarity threshold, the identification of the edge feature sample corresponding to the candidate coefficient is taken as a target identification.

[0139] Exemplarily, according to the plurality of feature recognition measurement points, the embodiments of the present application first compare these features with edge feature samples in an edge feature library, find the closest edge feature samples, combine the identifications of these edge feature samples to form an index vector, and then find a candidate device from a device archive through the index vector.

[0140] For the comparison of the features with the edge feature samples in the edge feature library, the embodiments of the present application use a retrieval model:

[0141]

[0142] wherein, PS is a similarity coefficient, SM ab is an element in the a-th row and b-th column of the edge feature matrix, Bm ab is an element in the a-th row and b-th column of the edge feature sample matrix, A is the total number of rows of the edge feature matrix, and B is the total number of columns of the edge feature matrix.

[0143] After the feature and the edge feature sample in the edge feature library are input into the retrieval model, a similarity coefficient is obtained, according to the similarity coefficient with the edge feature sample, the identification of the closest edge feature sample is found, and the identification is arranged in a certain order, and the index vector is obtained.

[0144] In step 104, in response to the confirmation instruction, the thermal imaging data is stored in the monitoring report of the candidate device.

[0145] Illustratively, the found candidate device is output to the maintenance personnel, and after the maintenance personnel confirms, the thermal imaging data can be stored in the monitoring report of the candidate device.

[0146] The embodiment of the application first acquires a first image and thermal imaging data, wherein the first image and the thermal imaging data are generated based on the same measurement point; then the first image is preprocessed, and a plurality of edge features are extracted from the preprocessed first image; then according to the plurality of edge features, an index vector is generated by using a retrieval model, and a candidate device is acquired from a device archive according to the index vector; finally, in response to a confirmation instruction, the thermal imaging data is stored in the monitoring report of the candidate device. The embodiment of the application extracts a plurality of features based on the edge of the image, generates an index vector by using a retrieval model, finds a candidate device according to the index vector, and finally stores the thermal imaging data in the monitoring report of the candidate device. Since the edge feature extraction operation is adopted, the amount of image data processing is greatly reduced, the identification is strong, the index vector is generated by using the retrieval model, the device archive is quickly searched, and some modern means such as computer technology can automatically perform the above operation without manual recording of the thermal imaging data, thereby improving the inspection efficiency.

[0147] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the application.

[0148] The following is the device embodiment of the application, and for details not described in detail, reference can be made to the corresponding method embodiments described above.

[0149] Figure 3 The structure schematic diagram of the infrared temperature measurement recording system provided by the embodiment of the application is shown, only the part related to the embodiment of the application is shown for convenience of description, and the details are as follows:

[0150] As Figure 3 shown, the infrared temperature measurement recording system comprises an infrared temperature measurement device 4 and a server 3, the infrared temperature measurement device 4 is in communication connection with the server 3;

[0151] The infrared temperature measuring device 4 collects a first image and thermal imaging data of the measurement point and sends the first image and the thermal imaging data to the server 3.

[0152] After receiving the first image and the thermal imaging data, the server 3 performs the steps of the above infrared temperature recording method.

[0153] In some scenarios, the server 3 includes a processor 300 and a memory 301, and the memory 301 stores a computer program 302 according to the infrared temperature recording method. When the processor 300 executes the above computer program 302, the steps of the infrared temperature recording method are realized.

[0154] In the above embodiments, the description of each embodiment has its own focus, and the parts not described or recorded in a certain embodiment can be referred to the related description of other embodiments.

[0155] Those of ordinary skill in the art can realize that the templates, units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art 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 application.

[0156] When the modules / units are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the processor executes the computer program, the steps of each of the above infrared temperature recording methods can be realized. The computer program includes computer program code, which can be in the form of source code, object code, executable file or some intermediate form, etc. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory, random access memory, electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0157] The above-described embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. An infrared thermography method, characterized in that, The method comprises the following steps: obtaining a first image and thermal imaging data, wherein the first image and the thermal imaging data are generated based on the same measurement point; preprocessing the first image and extracting a plurality of edge features from the preprocessed first image; generating an index vector using a retrieval model according to the plurality of edge features, comprising: obtaining an edge feature library, wherein the edge feature library includes a plurality of edge feature samples, and each edge feature sample is provided with an identifier; inputting the plurality of edge features into the retrieval model respectively, and obtaining a plurality of target identifiers through retrieval in the edge feature library by using the retrieval model, wherein the target identifier is the identifier of the edge feature sample closest to the plurality of edge features; arranging the plurality of target identifiers according to the size of the value, and constructing an index vector according to the arranged plurality of target identifiers; obtaining a candidate device from a device archive according to the index vector; in response to a confirmation instruction, storing the thermal imaging data in a monitoring report of the candidate device.

2. The infrared thermographic survey method of claim 1, wherein, The preprocessing of the first image comprises: desaturating the first image to obtain a grayscale image; counting the distribution intervals of a plurality of pixel grayscale values in the grayscale image; adjusting the plurality of pixel grayscale values according to the distribution intervals and the distribution quantities of the plurality of pixel grayscale values, so that the distribution intervals and the distribution quantities of the plurality of pixel grayscale values respectively meet the preset intervals and the preset distribution quantities.

3. The infrared thermographic survey method of claim 2, wherein, The adjusting of the plurality of pixel grayscale values according to the distribution intervals and the distribution quantities of the plurality of pixel grayscale values, so that the distribution intervals and the distribution quantities of the plurality of pixel grayscale values respectively meet the preset intervals and the preset distribution quantities, comprises: a distribution interval adjustment step: adjusting the plurality of pixel grayscale values according to a first formula, the preset intervals and the distribution intervals of the plurality of pixel grayscale values, wherein the first formula is: In the formula, pixel (rn) is the gray value after the adjustment of the nth iteration, pixel max is the maximum value in the plurality of pixel gray values, pixel min is the minimum value in the plurality of pixel gray values, v max is the upper limit value of the preset interval, v min is the lower limit value of the preset interval counting the number of the plurality of pixel grayscale values distributed in a plurality of subintervals to obtain a plurality of distribution quantities; if the deviation between the plurality of distribution quantities and the preset distribution quantities is greater than a threshold value, determining a target subinterval according to the plurality of distribution quantities and the preset distribution quantities, adjusting the plurality of pixel grayscale values according to a second formula and the target subinterval, and jumping to the distribution interval adjustment step, wherein the target subinterval is the interval with the largest deviation between the distribution quantity and the preset distribution quantity, and the second formula is: pixel (rn+1) = w p • DIF • (pixel (rn) - ITV) 2 + C In the formula, pixel (rn+1) is the pixel gray value after the adjustment of the nth iteration, w p is an adjustment coefficient, DIF is the difference between the target subinterval distribution quantity and the preset distribution quantity, ITV is the median value of the target subinterval, and C is a bias constant.

4. The infrared thermographic survey method of claim 1, wherein, extracting a plurality of edge features from the preprocessed first image, comprising: obtaining an edge extraction operator; obtaining a first edge feature map by convolution calculation according to the edge extraction operator and the preprocessed first image; performing edge cutting and segmentation on the first edge feature map to obtain a plurality of second edge feature maps; performing a pooling operation on the plurality of second edge feature maps according to a preset size to obtain a plurality of third feature maps; using the plurality of third feature maps as the plurality of edge features.

5. The infrared thermographic survey method of claim 4, wherein, The method comprises the following steps: obtaining a first image and thermal imaging data, wherein the first image and the thermal imaging data are generated based on the same measurement point; preprocessing the first image and extracting a plurality of edge features from the preprocessed first image; generating an index vector using a retrieval model according to the plurality of edge features, comprising: obtaining an edge feature library, wherein the edge feature library includes a plurality of edge feature samples, and each edge feature sample is provided with an identifier; inputting the plurality of edge features into the retrieval model respectively, and obtaining a plurality of target identifiers through retrieval in the edge feature library by using the retrieval model, wherein the target identifier is the identifier of the edge feature sample closest to the plurality of edge features; arranging the plurality of target identifiers according to the size of the value, and constructing an index vector according to the arranged plurality of target identifiers; obtaining a candidate device from a device archive according to the index vector; in response to a confirmation instruction, storing the thermal imaging data in a monitoring report of the candidate device. The preprocessing of the first image comprises: desaturating the first image to obtain a grayscale image; counting the distribution intervals of a plurality of pixel grayscale values in the grayscale image; adjusting the plurality of pixel grayscale values according to the distribution intervals and the distribution quantities of the plurality of pixel grayscale values, so that the distribution intervals and the distribution quantities of the plurality of pixel grayscale values respectively meet the preset intervals and the preset distribution quantities. The adjusting of the plurality of pixel grayscale values according to the distribution intervals and the distribution quantities of the plurality of pixel grayscale values, so that the distribution intervals and the distribution quantities of the plurality of pixel grayscale values respectively meet the preset intervals and the preset distribution quantities, comprises: a distribution interval adjustment step: adjusting the plurality of pixel grayscale values according to a first formula, the preset intervals and the distribution intervals of the plurality of pixel grayscale values, wherein the first formula is: counting the number of the plurality of pixel grayscale values distributed in a plurality of subintervals to obtain a plurality of distribution quantities; if the deviation between the plurality of distribution quantities and the preset distribution quantities is greater than a threshold value, determining a target subinterval according to the plurality of distribution quantities and the preset distribution quantities, adjusting the plurality of pixel grayscale values according to a second formula and the target subinterval, and jumping to the distribution interval adjustment step, wherein the target subinterval is the interval with the largest deviation between the distribution quantity and the preset distribution quantity, and the second formula is: extracting a plurality of edge features from the preprocessed first image, comprising: obtaining an edge extraction operator; obtaining a first edge feature map by convolution calculation according to the edge extraction operator and the preprocessed first image; performing edge cutting and segmentation on the first edge feature map to obtain a plurality of second edge feature maps; performing a pooling operation on the plurality of second edge feature maps according to a preset size to obtain a plurality of third feature maps; using the plurality of third feature maps as the plurality of edge features. The method comprises the following steps: obtaining an index indication, a horizontal edge operator matrix and a vertical edge operator matrix, wherein the horizontal edge operator matrix and the vertical edge operator matrix are of the same type; a matrix obtaining step of obtaining a first matrix from the preprocessed first image according to the index indication, wherein the first matrix is of the same type as the horizontal edge operator matrix and the vertical edge operator matrix; extracting horizontal edge features and vertical edge features according to a third formula, the first matrix, the horizontal edge operator matrix and the vertical edge operator matrix, wherein the third formula is: wherein EF is a horizontal edge feature or a vertical edge feature, PM mn is an element in the mth row and nth column of the first matrix, FM mn is an element in the mth row and nth column of the horizontal edge operator matrix or the vertical edge operator matrix, M is the total number of rows of the horizontal edge operator matrix or the vertical edge operator matrix, and N is the total number of columns of the horizontal edge operator matrix or the vertical edge operator matrix. adding the horizontal edge features and the vertical edge features into a horizontal feature matrix and a vertical feature matrix according to the index indication; if the index indication does not reach the end of the preprocessed first image, offsetting the index indication according to a preset order and jumping to the matrix obtaining step; otherwise, taking a sum matrix of the horizontal feature matrix and the vertical feature matrix as a sum feature matrix; setting values lower than an edge threshold in the sum feature matrix to zero to obtain the first edge feature map.

6. The infrared thermographic survey method of claim 4, wherein, the edge cropping and segmentation of the first edge feature map to obtain a plurality of second edge feature maps, comprising: obtaining a plurality of rows and columns with a number of continuous zero values exceeding a continuous threshold from the first edge feature map as a plurality of segmentation rows and a plurality of segmentation columns; segmenting the first edge feature map according to the segmentation rows and the plurality of segmentation columns to obtain a plurality of intermediate feature maps; deleting intermediate feature maps smaller than the preset size from the plurality of intermediate feature maps and taking the remaining plurality of intermediate feature maps as the plurality of second edge feature maps.

7. The infrared thermographic survey method of claim 1, wherein, the edge features and the edge feature samples are both matrices, and the retrieval model is: where PS is a similarity coefficient, SM ab is an element of the edge feature matrix in the ath row and the bth column, BM ab is an element of the edge feature sample matrix in the ath row and the bth column, A is the total number of rows of the edge feature matrix, and B is the total number of columns of the edge feature matrix.

8. The infrared thermographic survey method of claim 7, wherein, the inputting of the plurality of edge features into the retrieval model to obtain a plurality of target labels through retrieval of the edge feature library by the retrieval model, comprising: for each edge feature in the plurality of edge features, performing the following steps: inputting an edge feature and each edge feature sample in the plurality of edge feature samples into the retrieval model to obtain a plurality of similarity coefficients, wherein the plurality of similarity coefficients correspond to the plurality of edge feature samples; selecting a similarity coefficient with the largest value in the plurality of similarity coefficients as a candidate coefficient; if the candidate coefficient is greater than a similarity threshold, taking the label of the edge feature sample corresponding to the candidate coefficient as a target label.

9. An infrared thermographic recording system characterized by comprising: an infrared temperature measuring device and a server, wherein the infrared temperature measuring device is in communication connection with the server; the infrared temperature measuring device collects first images and thermal imaging data of a measurement point and sends the first images and the thermal imaging data to the server; after the server receives the first images and the thermal imaging data, performing the steps of the infrared temperature recording method according to any one of claims 1 to 8.

Images