Bridge steel box water accumulation infrared thermal imaging intelligent identification method and system

Abstract

The application discloses a bridge steel box ponding infrared thermal imaging intelligent identification method and system, which comprises the following steps: collecting an infrared image of a to-be-identified area and obtaining temperature values of the to-be-identified area; establishing a no-ponding infrared image data set of the to-be-identified area; calculating ponding identification indexes of each no-ponding infrared image in the no-ponding infrared image data set, calculating the average value and the standard deviation of the ponding identification indexes of all the no-ponding infrared images, and calculating control lines on a control chart by using a Shewhart control chart; searching the infrared image of the to-be-identified area in a sliding window manner and calculating a sliding window ponding identification index; constructing a ponding judgment index and judging whether the infrared image of the to-be-identified area has ponding or not; calculating a ponding area boundary binary segmentation threshold value for the infrared image determined as having ponding; calculating a binary segmentation pixel area of the infrared image of the to-be-identified area, and calculating a ponding area physical area according to the pixel area, the focal length and the shooting distance of an infrared imaging device.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of bridge steel box detection, in particular to a bridge steel box water accumulation infrared thermal imaging intelligent identification method and system. BACKGROUND

[0002] Steel box girder and steel box composite girder are bridge structure forms commonly used in highway and municipal bridges. The bridge deck is prone to damage under the repeated action of vehicle wheels. During the rainfall process, water seeps into the steel box from the bridge deck, causing water accumulation in the steel box, accelerating the corrosion of the steel box, and affecting the safe and durable operation of the bridge structure. The traditional inspection method is for inspectors to enter the box chamber for visual inspection. However, the majority of small and medium span steel box girders and steel box composite girders have a generally low beam height, and some bridges do not have maintenance manholes during construction, which makes it impossible to use the mode of periodic entry of personnel into the steel box for detection, restricting the timely and effective exploration of steel box water accumulation. During the operation of the bridge, it is found that there is a lot of water accumulation in some steel boxes, which poses a huge safety hazard. Therefore, it is necessary to explore new methods for detecting and identifying steel box water accumulation to solve the problem of being unable to enter the steel box for water accumulation detection.

[0003] Infrared thermal imaging technology can capture the temperature of the structure surface and detect structural defects remotely and non-contactly through temperature distribution anomalies. In recent years, it has become an important means of structural defect detection. The water accumulation area of the steel box will cause a significant difference in temperature from the adjacent area, and it is expected to identify water accumulation by collecting the temperature of the steel box bottom plate through infrared thermal imaging, realize the perspective exploration of the steel box water accumulation chamber outside, and overcome the disadvantages of traditional manual detection that requires entering the chamber.

[0004] However, the current steel box water accumulation infrared image intelligent screening and identification is low, mainly relying on manual identification, which is inefficient and subjective, and cannot meet the needs of large-area, batch, and instant detection. SUMMARY

[0005] To solve the problems in the prior art, the present application provides a bridge steel box water accumulation infrared thermal imaging intelligent identification method and system, which realizes the intelligent identification of steel box water accumulation infrared images and provides a new way for bridge steel box water accumulation detection and identification.

[0006] In a first aspect, the present application provides a bridge steel box water accumulation infrared thermal imaging intelligent identification method, which includes the following contents:

[0007] An infrared imaging device is used to collect infrared images of the area to be identified and obtain temperature values of the area to be identified;

[0008] An infrared image without water accumulation is selected for comparison to establish an infrared image data set without water accumulation;

[0009] Calculate the water identification index of each water-free infrared image in the water-free infrared image dataset; calculate the average and standard deviation of the water identification index of all water-free infrared images, and use the Shewhart control chart to calculate the control line UCL on the control chart;

[0010] Scan the infrared image of the to-be-identified region in a sliding window manner, and calculate the sliding window water identification index;

[0011] Establish a water identification index matrix of the infrared image of the to-be-identified region , represents the sliding window water identification index in the pth row and qth column of the water identification index matrix;

[0012] Construct a water determination index D:

[0013] ;

[0014] ;

[0015] represents the binary determination factor of the sliding window water identification index; k represents the starting column index of the continuous abnormal horizontal scanning of the sliding window; b represents the window width of the sliding window horizontal scanning determination anomaly; r and s represent the height and width of the sliding window respectively; u and v represent the horizontal and vertical moving steps of the sliding window scanning search respectively; n and m represent the number of rows and columns of the temperature matrix of the infrared image of the to-be-identified region;

[0016] If the water determination index D is greater than or equal to 1, it is determined that there is water in the infrared image of the to-be-identified region;

[0017] For the infrared image determined to have water, calculate the water region boundary binary segmentation threshold value;

[0018] Calculate the binary segmentation pixel area of the infrared image, and combine the focal length and shooting distance of the infrared imaging device to calculate the physical area of the water region.

[0019] In an optional implementation of the first aspect, after collecting the infrared image of the to-be-identified region, the infrared image of the to-be-identified region is preprocessed for noise reduction.

[0020] In an optional implementation of the first aspect, according to the historical results of manual inspection in the steel box, the water-free infrared image is selected for the to-be-identified region to establish a water-free infrared image dataset.

[0021] In an optional implementation of the first aspect, the specific steps of calculating the water identification index of each water-free infrared image in the water-free infrared image dataset are as follows:

[0022] Temperature matrix extracted from each waterless infrared image dataset. The average pixel temperature is calculated based on the pixel temperature value, and the standard deviation of the pixel temperature value is also calculated as an indicator for water accumulation detection. :

[0023] ;

[0024] This represents the pixel temperature value; The value represents the average temperature of the pixel; n and m represent the number of rows and columns of the temperature matrix T, respectively, i∈[1,n], j∈[1,m].

[0025] Specifically, the expression for the control line UCL on the control chart is as follows:

[0026] ;

[0027] , These represent the water accumulation identification metrics for all water-free infrared images in the water-free infrared image dataset. The mean and standard deviation; α represents the quantile of the standard normal distribution, and α represents the significance level. This indicates the number of samples in the infrared image dataset without water accumulation.

[0028] In one optional implementation of the first aspect, for infrared images determined to contain water accumulation, a genetic algorithm is used to determine the binarization segmentation threshold of the water accumulation region boundary; the specific steps are as follows:

[0029] Based on the highest and lowest temperature values ​​of the infrared image of the bridge steel box girder bottom plate, establish at equal intervals... Individual temperature , The objective function is defined as individual temperature. Average value of threshold segmentation boundary recognition index Set the initial threshold for the average value of the segmentation boundary recognition index;

[0030] Based on individual temperature Temperature matrix of infrared images of the steel box girder bottom plate of bridges determined to have water accumulation. Binarization is performed to obtain the binarized boundary matrix. ;

[0031] For the temperature matrix T of the infrared image of the steel box girder of the bridge determined to have water accumulation, in order to Set a sliding pane of size r×spx centered on the location, and calculate the corresponding water accumulation detection index for the sliding pane; calculate all... the average value of the sliding window recognition index of the position corresponding to the individual temperature as the threshold segmentation boundary recognition index average value of the individual temperature the average value of the corresponding threshold segmentation boundary recognition index ;

[0032] calculate the individual temperature of each individual in the temperature population the corresponding individual selection probability ;

[0033] calculate the maximum threshold segmentation boundary recognition index average value in the hth generation temperature population , ; if - the maximum threshold segmentation boundary recognition index average value of the hth generation temperature population is selected the corresponding individual temperature as the threshold segmentation boundary recognition index average value of the individual temperature; otherwise, the individual selection probability randomly select individuals from the hth generation temperature population with replacement, and combine them in pairs according to the index order to form several pairs of parents, each pair of parents performing crossover according to the individual selection probability generate 2 offspring, select a preset proportion of individuals in all offspring for mutation operation to change to a random temperature between the highest temperature value and the lowest temperature value, and generate the h+1th generation temperature population; preferably, the preset threshold is 0.001;

[0034] For the newly generated temperature population, repeat the above steps of binaryzation processing the temperature matrix of the bridge steel box bottom plate infrared image determined to have accumulated water, calculating the binaryzation boundary matrix, setting a sliding window with the position of the binaryzation boundary matrix element being 1 as the center, and taking the average value of the recognition index values of all sliding windows with the binaryzation boundary matrix element being 1 as the individual temperature corresponding to the threshold segmentation boundary recognition index average value, and further calculating the individual selection probability of the individual temperature, and determining whether the difference between the maximum threshold segmentation boundary recognition index average value of the newly generated temperature population and the maximum threshold segmentation boundary recognition index average value of the last generation temperature population is less than the preset threshold; until the difference between the maximum threshold segmentation boundary recognition index average value of the newly generated temperature population and the maximum threshold segmentation boundary recognition index average value of the last generation temperature population is less than the preset threshold.

[0035] select the individual temperature corresponding to the maximum threshold segmentation boundary recognition index average value in the current generation temperature population as the threshold segmentation boundary recognition index average value of the bridge steel box bottom plate infrared image.

[0036] Further, the temperature matrix of the bridge steel box bottom plate infrared image determined to have accumulated water is binaryzation processed to obtain a binaryzation boundary matrix The specific steps are as follows:

[0037] The temperature matrix of the bridge steel box bottom plate infrared image determined to have water accumulation The binary processing is performed to obtain a matrix A;

[0038] , ;

[0039] The binary boundary matrix B is calculated according to the following formula:

[0040] ;

[0041] , ;

[0042] , ;

[0043] , , , are elements of the matrices A, B, V, and H, respectively, and the symbol represents the union of each element of the two matrices, and the number of elements of the matrix B is the same as the number of elements of the matrices V and H.

[0044] In an optional implementation of the first aspect, the expression of the physical area of the water accumulation region is as follows:

[0045] ;

[0046] represents the physical area of the water accumulation region; represents the focal length of the infrared imaging device; represents the shooting distance of the infrared imaging device; represents the binary segmentation pixel area of the infrared image.

[0047] In a second aspect, the application provides a system for executing the bridge steel box water accumulation infrared thermal imaging intelligent identification method, comprising:

[0048] an infrared imaging device, configured to capture an infrared image of a to-be-identified region and obtain a temperature value of the to-be-identified region;

[0049] a preprocessing module, configured to preprocess the infrared image of the to-be-identified region;

[0050] The non-ponding infrared image processing module selects non-ponding infrared images by comparison, establishes a non-ponding infrared image dataset, and calculates the ponding identification index of each non-ponding infrared image in the non-ponding infrared image dataset; the average value and the standard deviation of the ponding identification index of all non-ponding infrared images are calculated, and the Shewhart control chart is used to calculate the control line UCL on the control chart;

[0051] The to-be-identified area ponding determination module scans the to-be-identified area infrared image in a sliding window manner, calculates the ponding identification index of each sliding window, and establishes a ponding identification index matrix of the to-be-identified area infrared image; a ponding determination index is constructed, and whether the to-be-identified area infrared image has ponding is determined according to the ponding determination index;

[0052] The binary segmentation threshold calculation module calculates the binary segmentation threshold of the ponding area boundary for the infrared image determined to have ponding.

[0053] The ponding area physical area calculation module calculates the physical area of the ponding area according to the focal length, shooting distance of the infrared imaging device, and binary segmentation pixel area of the infrared image.

[0054] Compared with the prior art, the present application has the following beneficial effects:

[0055] 1. The present application establishes a ponding identification index and a determination index by directly extracting the temperature of an infrared image, and automatically screens ponding images based on quantitative calculation in the whole process. The image batch collected in a large area can be imported once to output the screening result in real time, so that the identification efficiency and objectivity are effectively guaranteed, and the intelligent upgrading promotes the industry-level application of infrared thermal imaging detection of ponding in steel box chambers.

[0056] 2. The present application uses a genetic algorithm to optimize the binary segmentation threshold for the infrared image determined to have ponding, automatically locks the ponding boundary, calculates the area of the ponding area, realizes the accurate inversion identification of the ponding area of the infrared image, and significantly improves the identification accuracy of the ponding area. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 A flowchart of a bridge steel box ponding infrared thermal imaging intelligent identification method provided for an embodiment of the present application;

[0058] Figure 2 A Shewhart control chart provided for an embodiment of the present application;

[0059] FIG. 3(a) is a schematic diagram of the result of pre-processing of a to-be-identified infrared image Photo 1 by Gaussian filter noise reduction provided for an embodiment of the present application;

[0060] FIG. 3(b) is a schematic diagram of the result of pre-processing of a to-be-identified infrared image Photo 2 by Gaussian filter noise reduction provided for an embodiment of the present application;

[0061] Fig. 3(c) is a schematic diagram of a result of the Gaussian filter denoising pre-processing of an infrared image Photo 3 to be identified according to an embodiment of the present application;

[0062] Fig. 3(d) is a schematic diagram of a result of the Gaussian filter denoising pre-processing of an infrared image Photo 4 to be identified according to an embodiment of the present application;

[0063] Fig. 3(e) is a schematic diagram of a result of the Gaussian filter denoising pre-processing of an infrared image Photo 5 to be identified according to an embodiment of the present application;

[0064] Fig. 4(a) is a Shewhart control chart of an infrared image Photo 1 to be identified according to an embodiment of the present application;

[0065] Fig. 4(b) is a Shewhart control chart of an infrared image Photo 2 to be identified according to an embodiment of the present application;

[0066] Fig. 4(c) is a Shewhart control chart of an infrared image Photo 3 to be identified according to an embodiment of the present application;

[0067] Fig. 4(d) is a Shewhart control chart of an infrared image Photo 4 to be identified according to an embodiment of the present application;

[0068] Fig. 4(e) is a Shewhart control chart of an infrared image Photo 5 to be identified according to an embodiment of the present application;

[0069] Fig. 5(a) is a schematic diagram of a binaryzation segmentation result of a water-accumulation area boundary of an infrared image photo 4 to be identified according to an embodiment of the present application;

[0070] Fig. 5(b) is a schematic diagram of a binaryzation segmentation result of a water-accumulation area boundary of an infrared image photo 5 to be identified according to an embodiment of the present application. DETAILED DESCRIPTION

[0071] In order to make the objects, technical solutions, and advantages of the present application clearer, the following will further describe the present application with reference to the accompanying drawings.

[0072] It should be noted that the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0073] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. The singular forms “a,” “the,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0074] Firstly, such as Figure 1 As shown, this application provides an infrared thermal imaging intelligent identification method for water accumulation in bridge steel box girders, including the following:

[0075] S1: Use an infrared imaging device to acquire an infrared image of the area to be identified and obtain the temperature value of the area to be identified; in this application, the area to be identified is mainly the bottom plate of the steel box girder of the bridge to be identified; preferably, use an infrared thermal imager or a drone with an infrared camera to photograph the area to be identified;

[0076] For example, step S1 further includes a step of preprocessing the acquired infrared image of the region to be identified, wherein the preprocessing includes, but is not limited to, using Gaussian filtering to reduce noise in the infrared image.

[0077] S2: Select infrared images of the corresponding area without water accumulation to establish a dataset of infrared images without water accumulation; for example, the dataset can be constructed by comparing the selected infrared images of the area to be identified with the historical results of manual inspections inside the steel box.

[0078] S3: Calculate the water accumulation identification index for each water-free infrared image in the water-free infrared image dataset;

[0079] The specific steps are as follows:

[0080] S3-1: Extracting the temperature matrix of each waterless infrared image from the waterless infrared image dataset. , These are the elements of the temperature matrix T, i.e., the pixel temperature values;

[0081] S3-2: Calculate the average pixel temperature based on the pixel temperature value, and calculate the standard deviation of the pixel temperature, using it as an indicator for water accumulation detection. :

[0082] ;

[0083] In the formula, This represents the pixel temperature value; The value represents the average temperature of the pixel; n and m represent the number of rows and columns of the temperature matrix T, respectively, i∈[1,n], j∈[1,m].

[0084] S4: Calculate the water accumulation identification index for all water-free infrared images in the water-free infrared image dataset. The mean and standard deviation are used to determine the control chart using Shewhart control chart theory, and the control line UCL on the control chart is calculated.

[0085] Specifically, the expression for the control line UCL on the control chart is as follows:

[0086] ;

[0087] In the formula, , These represent the water accumulation identification metrics for all water-free infrared images in the water-free infrared image dataset. The mean and standard deviation; The quantiles represent the standard normal distribution, which are correlated with the significance level α and are used to determine the width of the control limits, reflecting the tolerance for process fluctuations, etc. This indicates the number of samples in the infrared image dataset without water accumulation.

[0088] The control chart uses water accumulation as an indicator. The standard deviation is plotted on the ordinate, and the sample number of the waterless infrared image dataset is plotted on the x-axis, such as... Figure 2 As shown, the red line represents the upper control line (UCL).

[0089] S5: Scan the infrared image of the steel box girder bottom plate of the bridge to be identified line by line using the sliding pane method, and calculate the water accumulation identification index of the sliding pane; based on the water accumulation identification index of the sliding pane, establish the water accumulation identification index matrix of the infrared image of the steel box girder bottom plate of the bridge to be identified.

[0090] The specific steps are as follows:

[0091] S5-1: Define a rectangular window with a fixed size of r×s px (pixels) as the sliding pane for scanning search; r represents the height of the sliding pane, and s represents the width of the sliding pane;

[0092] S5-2: Starting from the upper left of the infrared image of the steel box girder of the bridge to be identified, the infrared image of the steel box girder of the bridge to be identified is scanned line by line using an r×s px sliding pane method. During the line-by-line scanning search, the horizontal movement step of the r×s px sliding pane is u pixels. After each line is scanned, the sliding pane moves vertically down v pixels to perform the second line scan search, and so on, until the lower right corner of the sliding pane covers the lower right corner of the image, thus completing the entire scan of the image.

[0093] S5-3: Calculate the water accumulation identification index of the sliding window, that is, the water accumulation identification index of the area enclosed by each rectangular window in the infrared image of the bottom plate of the steel box of the bridge to be identified. The calculation method of the water accumulation identification index of the sliding window is the same as step S3-2.

[0094] S5-4: Based on the sliding window water identification index, the water identification index matrix of the infrared image of the bridge steel box bottom plate to be identified is established:

[0095] ;

[0096] p, q represent the pth row and qth column of the water identification index matrix, respectively; , ; represents the sliding window water identification index in the pth row and qth column of the water identification index matrix; the symbol represents rounding down;

[0097] S6: Construct the water determination index D to determine whether there is a continuous abnormal window in the infrared image of the bridge steel box bottom plate to be identified;

[0098] The specific steps are as follows:

[0099] S6-1: Construct the water determination index D:

[0100] ;

[0101] ;

[0102] represents the binary determination factor of the sliding window water identification index; k represents the sliding window continuous abnormal horizontal scanning starting column index, ; b represents the window width of the sliding window horizontal scanning determination abnormality, which needs to be set according to the actual recognition accuracy requirement;

[0103] S6-2: If the water determination index D is greater than or equal to 1, it indicates that there are continuous b elements in the recognition index matrix F of the infrared image of the bridge steel box bottom plate to be identified, which exceed the control line UCL on the control chart, and it is determined that there is water in the infrared image of the bridge steel box bottom plate to be identified.

[0104] S7: For the bridge steel box bottom plate infrared image determined to have water, calculate the water area boundary binary segmentation threshold; preferably, use genetic algorithm to find the water area boundary binary segmentation threshold;

[0105] The specific steps are as follows:

[0106] S7-1: According to the highest temperature value and the lowest temperature value of the bridge steel box bottom plate infrared image, establish individual temperature , ; define the threshold segmentation boundary identification index average value of the individual temperature , set the initial threshold segmentation boundary identification index average value; conventionally, the initial threshold segmentation boundary identification index average value is set to 0;

[0107] S7-2: According to the individual temperature , the temperature matrix of the bridge steel box bottom plate infrared image determined to have water accumulation , the matrix A is obtained by performing binaryzation processing; the expression of the matrix A is as follows:

[0108] , ;

[0109] Calculate the binaryzation boundary matrix ;

[0110] ;

[0111] , ;

[0112] , ;

[0113] , , , , respectively, the elements of the matrix A, B, V, and H, the symbol represents the union of the elements of the two matrices; the number of elements of the matrix B is the same as the number of elements of the matrix V and H;

[0114] S7-3: For the temperature matrix of the bridge steel box bottom plate infrared image determined to have water accumulation , the position of is taken as the center to set a sliding window (with a size of r x s px), and the corresponding sliding window water accumulation identification index is calculated; the calculation process is referred to step S5-3; the average value of all sliding window water accumulation identification indexes is calculated, and the average value is taken as the individual temperature corresponding to the threshold segmentation boundary identification index average value ;

[0115] S7-4: Calculate the individual selection probability corresponding to each individual temperature ;

[0116] S7-5: Calculate the maximum threshold segmentation boundary identification index average value in the hth generation population , ; if - ​If the difference between the maximum threshold segmentation boundary identification index average value of the current generation population and the maximum threshold segmentation boundary identification index average value of the last generation population (i.e. the difference between the maximum threshold segmentation boundary identification index average value of the current generation population and the maximum threshold segmentation boundary identification index average value of the last generation population) is less than a preset threshold, step S7-6 is skipped and step S7-7 is executed, otherwise step S7-6 is executed; for example, the preset threshold can be 0.001;

[0117] S7-6: According to the individual selection probability randomly selecting individual temperatures from the parent (i.e. the hth generation) temperature population, combining the individual temperatures in index order to form a plurality of pairs of parents, and performing crossover on each pair of parents according to the individual selection probability to generate two offspring, selecting a preset proportion of individuals in all offspring for mutation operation to change to a random temperature between the highest temperature value and the lowest temperature value, and generating an (h+1)th generation temperature population; returning to step S7-2.

[0118] For example, each pair of parents (for example, t1 and t2) can perform crossover with a probability of 0.8, and the specific operation steps are as follows: generating a random value between [0, 1] by a random number generator, if the value is less than or equal to 0.8, the crossover operation on the current pair of parent individuals is triggered; otherwise, the crossover operation is skipped. If the crossover operation is triggered, a random coefficient β (β ∈ [0, 1]) is generated, and two offspring are generated by the following formula: t child1 = βt1 + (1-β)t2, t child2 = βt2 + (1-β)t1. If the crossover operation is not triggered, the parent individuals t1 and t2 are directly reserved as offspring individuals; 10% of the individuals in all offspring are selected for mutation operation.

[0119] S7-7: Selecting the individual temperature corresponding to the maximum threshold segmentation boundary identification index average value in the current generation temperature population as the accumulated water area boundary binary segmentation threshold of the bridge steel box bottom plate infrared image.

[0120] S8: According to the accumulated water area boundary binary segmentation threshold, calculating the pixel area of the bridge steel box bottom plate infrared image with a binary value of 1 (i.e. ), that is, the infrared image binary segmentation pixel area; according to the infrared image binary segmentation pixel area, the focal length and the shooting distance of the infrared imaging device, calculating the physical area of the accumulated water area

[0121]

[0122] represents the focal length of the infrared imaging device; represents the shooting distance of the infrared imaging device; ​​represents the pixel area of the infrared image binaryzation segmentation, that is, the pixel area of the infrared image binaryzation value of 1; the formula converts the pixel size into the physical size by the ratio of the focal length of the infrared imaging device to the shooting distance.

[0123] In a second aspect, a system for performing the bridge steel box ponding infrared thermal imaging intelligent identification method is provided, comprising:

[0124] an infrared imaging device for shooting an infrared image of a to-be-identified region to obtain a temperature value of the to-be-identified region;

[0125] a preprocessing module for preprocessing the infrared image of the to-be-identified region;

[0126] a non-ponding infrared image processing module for selecting a non-ponding infrared image from the to-be-identified region, establishing a non-ponding infrared image dataset, and calculating a ponding identification index of each non-ponding infrared image in the non-ponding infrared image dataset; calculating the average value and the standard deviation of the ponding identification index of all non-ponding infrared images, and calculating a control line UCL on a Shewhart control chart;

[0127] a to-be-identified region ponding determination module for searching the infrared image of the to-be-identified region in a sliding window manner, calculating a ponding identification index of each sliding window, and establishing a ponding identification index matrix of the infrared image of the to-be-identified region; constructing a ponding determination index, and determining whether the infrared image of the to-be-identified region exists ponding according to the ponding determination index;

[0128] a binaryzation segmentation threshold calculation module for calculating a binaryzation segmentation threshold of a ponding region boundary for the infrared image determined to exist ponding;

[0129] a ponding region physical area calculation module for calculating a binaryzation segmentation pixel area of the infrared image according to the binaryzation segmentation threshold of the ponding region boundary, and calculating a ponding region physical area according to the focal length of the infrared imaging device, the shooting distance, and the binaryzation segmentation pixel area of the infrared image.

[0130] It should be understood that the division of each processing unit in the above system is only a logical functional division, and all or part of the processing units can be integrated into one physical entity, or can be physically separated. In addition, the processing units in the system can be realized in the form of processor calling software; for example, the system includes a processor, the processor is connected with a memory, the memory stores instructions, and the processor calls the instructions stored in the memory to realize the above method or the function of each processing unit of the system, wherein the processor is a general-purpose processor, such as a central processing unit or a microprocessor, and the memory is a memory in the system or a memory outside the system.

[0131] Thirdly, this application also provides a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, implement the methods described in the above embodiments.

[0132] For example, the above can be achieved Figure 1 The steps of the infrared thermal imaging intelligent identification method for water accumulation in bridge steel box shown are illustrated.

[0133] A computer-readable storage medium can be a tangible device capable of holding and storing instructions used by an instruction execution device. A computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof.

[0134] The computer program instructions used to perform the operations of this application may be source code or object code written in any combination of one or more programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server.

[0135] The feasibility and effectiveness of this application will be explained in detail below with a simulation example.

[0136] S1: Infrared images of the bridge steel box girder base plate in the area to be identified are captured using an infrared thermal imager to obtain the temperature value of the steel box girder base plate in the area to be identified. Gaussian filtering is then used to preprocess the infrared image of the area to be identified for noise reduction. The Gaussian convolution kernel size is 9×9, and the standard deviation is 1.6. Figures 3(a)-3(e) The image shown is a schematic diagram of the result after Gaussian filtering and noise reduction processing of five infrared images (photo1-photo5) of the steel box girder of the bridge area to be identified, taken by an infrared thermal imager.

[0137] S2: Based on the results of manual inspections inside the steel box, infrared images of areas without water accumulation were selected and a dataset of infrared images without water accumulation was established, with a total sample size of 16,000 images.

[0138] S3: Extract the temperature matrix T from the water-free infrared image and calculate the average temperature of each pixel. Calculate the standard deviation of pixel temperature as an indicator for water accumulation detection. .

[0139] S4: Calculate all image recognition metrics in the waterless infrared image dataset. average and standard deviation Using Shewhart control chart theory, the control line UCL on the control chart is determined to be 0.258. Figures 4(a)-4(e)The Shewhart control chart corresponding to photo1-photo5 of the embodiment is shown, wherein the red line represents the upper control line UCL.

[0140] S5: For the infrared image of the bridge steel box bottom plate to be identified, the temperature matrix has a size of 300x300, and the image is scanned row by row starting from the top left corner in a sliding window manner of 100x100 px, the horizontal moving step is 1 pixel, after each row scan is completed, the vertical direction is moved down by 50 pixels for the second row scan, until the entire image scan is completed; the identification index of each sliding window is calculated ; the identification index matrix F of the to-be-identified infrared image is established; =5, =201, that is, F is a matrix of 5 rows and 201 columns.

[0141] S6: Take b=10, calculate the water accumulation determination index, if D≥1, it represents that there are 10 continuous elements exceeding the upper control line UCL in the identification index matrix F in the horizontal direction, and it is determined that there is water accumulation in the to-be-identified image. In photo1-photo3, S=0<1, indicating that there is no water accumulation in the image; in photo4, S=244>1; in photo5, S=183>1, it is determined that there is water accumulation in the image.

[0142] S7: The genetic algorithm is used to determine the segmentation optimization threshold of the water accumulation area, and the specific steps are as follows:

[0143] S7-1: According to the maximum temperature value and the minimum temperature value of the image, 50 individual temperatures (g=1,2,……,50) are established at equal intervals; the initial threshold segmentation boundary identification index average value is set to 0;

[0144] S7-2: According to the individual temperature , the temperature matrix T of the bridge steel box bottom plate infrared image determined to have water accumulation is binarized to obtain the matrix , and the binarization boundary matrix is further calculated.

[0145] S7-3: For the temperature matrix T of the bridge steel box bottom plate infrared image determined to have water accumulation, the sliding window is set with the position of as the center, the sliding window water accumulation identification index is calculated, the average value of the sliding window water accumulation identification indexes of all is calculated, which represents the threshold segmentation boundary identification index average value corresponding to the individual temperature .

[0146] S7-4: The individual selection probability corresponding to the individual temperature is calculated.

[0147] S7-5: Calculate the average value of the maximum threshold segmentation boundary identification index in the hth generation population , if - is less than 0.001, skip step S7-6 and execute step S7-7, otherwise execute step S7-6;

[0148] S7-6: Randomly select 50 individual temperatures from the parent temperature population according to the individual selection probability, combine them in pairs according to the index order to form several pairs of parents, and select the individual temperature of each pair of parents according to the probability Perform crossover to generate 2 offspring, select a preset proportion of individuals in all offspring for mutation operation, change to a random temperature between the highest temperature value and the lowest temperature value, and generate the h+1th generation temperature population; return to step S7-2;

[0149] S7-7: Use the individual temperature corresponding to the average value of the maximum threshold segmentation boundary identification index in the current generation population as the water accumulation area boundary binary segmentation threshold of the image. For photo4, the segmentation threshold is 22.85℃; for photo5, the segmentation threshold is 23.35℃. As shown in Fig. 5(a) and Fig. 5(b), the water accumulation area boundary binary segmentation results of photo4 and photo5 determined according to the water accumulation area boundary binary segmentation threshold are shown.

[0150] S8: According to the water accumulation area boundary binary segmentation result, calculate the infrared image binary segmentation pixel area of photo4 and photo5 , ; According to the focal length , shooting distance of the infrared thermal imager, the physical area of the water accumulation area of photo4 and photo5 is calculated as 0.3290m 2 and 0.3310m 2 , respectively.

[0151] The above examples are only for illustrating the technical concept and characteristics of the present application, and the purpose is to enable those skilled in the art to understand the content of the present application and implement it, and cannot limit the protection scope of the present application. For ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should also be considered as the protection scope of the present application.

Claims

1. A bridge steel box ponding infrared thermal imaging intelligent identification method, characterized in that, The method comprises the following steps: An infrared imaging device is used to collect an infrared image of a region to be identified, and obtain a temperature value of the region to be identified; Non-water-logged infrared images are selected by comparison, and a non-water-logged infrared image dataset is established; extracting a temperature matrix of each non-ponding infrared image from a non-ponding infrared image data set , calculating a pixel temperature average value according to the pixel temperature value, and calculating a pixel temperature value standard deviation as a ponding recognition index : ; wherein, represents a pixel point temperature value; represents a pixel point temperature average value; n, m represent the row and column numbers of the temperature matrix of the infrared image of the region to be identified, i∈[1,n], j∈[1,m]; The average value and the standard deviation of the water-logging identification index of all non-water-logged infrared images are calculated, and a Shewhart control chart is used to calculate a control line UCL on the control chart: ; wherein , respectively represent the average value and the standard deviation of the water identification index of all water-free infrared images in the water-free infrared image dataset represents the quantile of the standard normal distribution, and a represents the significance level represents the sample number of the water-free infrared image dataset​ The infrared image of the region to be identified is scanned in a sliding window manner, and the water-logging identification index of the sliding window is calculated; An accumulated water identification index matrix of an infrared image of a region to be identified is established , represents a sliding window accumulated water identification index in the pth row and qth column of the accumulated water identification index matrix A water-logging determination index D is constructed: ； ； a binary decision factor representing the sliding window pane water accumulation identification index; k represents the sliding window pane continuous abnormal horizontal scanning starting column index; b represents the window width of the sliding window pane horizontal scanning decision abnormality; r, s respectively represent the height, width of the sliding window pane; u, v respectively represent the horizontal moving step, vertical moving step of the sliding window pane scanning search; n, m represent the row, column number of the temperature matrix of the infrared image of the region to be identified; If the water-logging determination index D is greater than or equal to 1, it is determined that the infrared image of the region to be identified contains water-logging; For the infrared image determined to contain water-logging, the water-logging region boundary binarization segmentation threshold value is calculated; The binarization segmentation pixel area of the infrared image is calculated, and the focal length and the shooting distance of the infrared imaging device are combined to calculate the physical area of the water-logging region.

2. The bridge steel box ponding infrared thermal imaging intelligent identification method according to claim 1, characterized in that, After the infrared image of the region to be identified is collected, the infrared image of the region to be identified is preprocessed.

3. The bridge steel box ponding infrared thermal imaging intelligent identification method according to claim 1, characterized in that, According to the historical results of manual inspection in the steel box, non-water-logged infrared images are selected by comparison, and a non-water-logged infrared image dataset is established.

4. The bridge steel box ponding infrared thermal imaging intelligent identification method according to claim 1, characterized in that, For the infrared image determined to contain water-logging, a genetic algorithm is used to determine the water-logging region boundary binarization segmentation threshold value; the specific steps are as follows: According to the highest value and the lowest value of the temperature of the bridge steel box bottom plate infrared image, an equal interval is established individual temperature , ; defining a target function as a temperature of the individual a threshold segmentation boundary identification index average value setting an initial threshold segmentation boundary identification index average value; According to the individual temperature The temperature matrix of the infrared image of the bridge steel box bottom plate determined as having water accumulation The binary boundary matrix is obtained by performing a binaryzation process ; For the temperature matrix T of the infrared image of the bridge steel box bottom plate judged to have water accumulation, an rxs px sliding window is set with the position of The average value of the sliding window water accumulation identification index of all The average value of the sliding window water accumulation identification index of all The average value of the sliding window water accumulation identification index of all The average value of the sliding window water accumulation identification index of all Computing the temperature of each individual in the temperature population Corresponding individual selection probability ; calculating the maximum threshold segmentation boundary identification index average value in the hth generation temperature population , ; if - the maximum threshold segmentation boundary identification index average value in the hth generation temperature population is selected as the binary threshold value of the accumulated water area boundary of the infrared image of the bridge steel box bottom plate; otherwise, the individual temperature corresponding to the individual selection probability is selected as the binary threshold value of the accumulated water area boundary of the infrared image of the bridge steel box bottom plate; otherwise, the individual temperature corresponding to the individual selection probability is selected as the binary threshold value of the accumulated water area boundary of the infrared image of the bridge steel box bottom plate; otherwise, the individual temperature corresponding to the individual selection probability is selected as the binary threshold value of the accumulated water area boundary of the infrared image of the bridge steel box bottom plate; otherwise, the individual temperature corresponding to the individual selection probability is selected as the binary threshold value of the accumulated water area boundary of the infrared image of the bridge steel box bottom plate; otherwise, the individual temperature corresponding to the individual selection probability For the newly generated temperature population, the above-mentioned steps of binarizing the temperature matrix of the infrared image of the bridge steel box bottom plate determined to contain water-logging are repeated, the binarization boundary matrix is calculated, the position with an element of 1 in the binarization boundary matrix is taken as the center to set a sliding window, the average value of the sliding window identification index values of all the elements with 1 in the binarization boundary matrix is taken as the threshold segmentation boundary identification index average value corresponding to the individual temperature, and then the individual selection probability of the individual temperature is calculated; whether the difference between the maximum threshold segmentation boundary identification index average value of the newly generated temperature population and the maximum threshold segmentation boundary identification index average value of the last generation temperature population is less than a preset threshold value is judged; until the difference between the maximum threshold segmentation boundary identification index average value of the newly generated temperature population and the maximum threshold segmentation boundary identification index average value of the last generation temperature population is less than the preset threshold value; The individual temperature corresponding to the maximum threshold segmentation boundary identification index average value in the current generation temperature population is selected as the water-logging region boundary binarization segmentation threshold value of the infrared image of the bridge steel box bottom plate.

5. The bridge steel box ponding infrared thermal imaging intelligent identification method according to claim 4, characterized in that, The temperature matrix of the infrared image of the bridge steel box bottom plate determined as having water accumulation The binarization processing is performed to obtain a binarization boundary matrix The specific steps are as follows: Temperature matrix of an infrared image of a bridge steel box bottom plate determined to have water accumulation Binary processing is performed to obtain a matrix A; ， ； The binarization boundary matrix B is calculated according to the following formula: ； ， ； ， ； , , , are elements of matrices A, B, V, H, respectively, and the notation denotes taking the union of the elements of the two matrices, and the number of elements of matrix B is the same as the number of elements of matrices V, H.

6. The bridge steel box ponding infrared thermal imaging intelligent identification method according to claim 4, characterized in that, The preset threshold value is 0.

001.

7. The bridge steel box ponding infrared thermal imaging intelligent identification method according to claim 1, characterized in that, The expression of the physical area of the water-logging region is as follows: ； represents a physical area of a water accumulation region; represents a focal length of an infrared imaging device; represents a shooting distance of an infrared imaging device; represents a binaryzation segmented pixel area of an infrared image.

8. A system for performing the method of intelligent identification of accumulated water in a bridge steel box by infrared thermography according to any one of claims 1 to 7, characterized in that, The method comprises the following steps: An infrared imaging device is used to collect an infrared image of a region to be identified, and obtain a temperature value of the region to be identified; A preprocessing module is used to preprocess the infrared image of the region to be identified; A non-water-logged infrared image processing module is used to select non-water-logged infrared images by comparison, establish a non-water-logged infrared image dataset, and calculate the water-logging identification index of each non-water-logged infrared image in the non-water-logged infrared image dataset; the average value and the standard deviation of the water-logging identification index of all non-water-logged infrared images are calculated, and a Shewhart control chart is used to calculate a control line UCL on the control chart; The waterlogging area to be identified determination module adopts a sliding window mode to scan the infrared image of the area to be identified, calculates waterlogging identification indexes of each sliding window, and establishes a waterlogging identification index matrix of the infrared image of the area to be identified; a waterlogging determination index is constructed, and whether the infrared image of the area to be identified has waterlogging is determined according to the waterlogging determination index; The binary segmentation threshold value calculation module calculates a binary segmentation threshold value of a waterlogging area boundary for the infrared image determined to have waterlogging; The waterlogging area physical area calculation module calculates a waterlogging area physical area according to a focal length of the infrared imaging device, a shooting distance and a binary segmentation pixel area of the infrared image.

Images