An Identification and Segmentation Method for Insulator Infrared Images Based on Edge Features

Abstract

The present invention discloses a method for identifying and segmenting infrared images of insulators based on edge features, which comprises the following steps: Step 1, acquiring images and making an original data set; Step 2, preprocessing the images; Step 3, identifying according to insulator features; Step 4, extracting image features to obtain precise positioning and segmentation of individual insulators. By adopting the above technical solution, feature simplification is carried out by threshold binarization, and then image correction based on contour extraction is used to obtain the corrected infrared insulator image. After obtaining the corrected infrared insulator image, through one-dimensional pixel scanning, the boundary points of the insulator are calibrated, and then the distribution characteristics of the insulator boundary points are obtained through multi-dimensional scanning, so as to realize the extraction and segmentation of the insulator string. This method uses multi-dimensional scanning to extract the distribution characteristics of the boundary points of the insulators in the infrared image, making the identification and segmentation of the insulators in the infrared image have better portability, ease of use and are easy to expand functions.

Description

Technical Field

[0001] The present invention relates to the technical field of digital image processing, and particularly to a method for identifying and segmenting infrared images of insulators based on edge features. Background Art

[0002] The infrared image of an insulator can convert the infrared radiation distribution in space into an RGB image with corresponding color distribution. Since the temperature of the insulator string is higher than that of the surrounding environment, infrared sampling of the insulator is one of the important methods for identifying the insulator string and segmenting individual insulators. Deteriorated insulators pose a great potential hazard to the stability of power grid operation, and abnormal temperature can be used to judge whether the insulator has a high probability of deterioration. The identification and segmentation of infrared images of insulators are the key steps for automatically detecting whether the insulators are deteriorated. Therefore, constructing an accurate method for identifying and segmenting infrared images of insulators is of great significance and has important application value in the detection of deteriorated insulators.

[0003] There are not many achievements in the prior art in this regard. For example, a Chinese patent was published with the application number CN201910062932.1 and the title of "A Method for Identifying and Locating Insulators Based on Shape Features and Image Segmentation". The specific steps of this method are as follows: (1) Preprocess the original insulator image to reduce noise interference and improve image contrast; (2) Introduce an adaptive particle swarm optimization algorithm and combine it with the maximum inter-class variance method to segment the insulator image; (3) Form a binary image through morphological filtering and area filtering, extract the skeleton of the binary image, and perform initial positioning of the insulator area according to the Hough transform result; (4) Rely on the shape features of the insulator to achieve precise positioning of the insulator. The present invention makes full use of the prior knowledge of the shape features of the insulator and the technical methods of image processing to realize the automatic identification and positioning of the insulator string, laying a foundation for subsequent insulator fault diagnosis. The invention is practical and greatly improves the accuracy and efficiency of positioning.

[0004] The above-disclosed technical solution preprocesses the original insulator image to reduce noise interference and improve image contrast; introduces an adaptive particle swarm optimization algorithm and combines it with the maximum inter-class variance method to segment the insulator image; forms a binary image through morphological filtering and area filtering, extracts the skeleton of the binary image, and performs initial positioning of the insulator area according to the Hough transform result; relies on the shape features of the insulator to achieve precise positioning of the insulator.

[0005] However, it mainly relies on the Hough transform to locate the insulators and on the overall shape features of the insulators for positioning. Since it is necessary to comprehensively identify and locate the overall shape features, in order to ensure the accuracy of identification and positioning, a large amount of image data needs to be obtained, compared and analyzed. During the entire comparison and analysis process, the computational workload is very large and it takes a long time, which has a certain limitation on the automation performance. Summary of the Invention

[0006] According to the deficiencies of the prior art, the present invention proposes a method for identifying and segmenting infrared images of insulators based on edge features. After image correction based on contour extraction, boundary point calibration of the insulators is performed, the distribution characteristics of boundary points are obtained through multi-dimensional scanning, and then according to the shape characteristics of the insulators themselves, the insulator string is quickly and accurately identified and the insulators are segmented.

[0007] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0008] A method for identifying and segmenting infrared images of insulators based on edge features, comprising the following steps:

[0009] Step 1: Collect images and create an original data set;

[0010] Step 2: Preprocess the images;

[0011] Step 2.1: Denoise the original data set;

[0012] Step 2.2: Perform threshold binaryzation using image features;

[0013] Step 2.3: Correct the image based on contour extraction;

[0014] Step 2.4: Calibrate edge points by traversal method using the corrected image;

[0015] Step 3: Identify according to insulator features;

[0016] Step 4: Extract the image features to obtain precise positioning and segmentation of individual insulators.

[0017] Preferably, in Step 1, the images are collected by an infrared camera. In Step 2.1, the original data set is denoised by Gaussian filtering. In Step 2.3, the extracted contour image is corrected by one-dimensional pixel traversal.

[0018] Preferably, the edge point calibration by traversal method includes calibrating the corrected contour image, and the calibrated points are points within the insulator region on the image, and the previous pixel point and the next pixel point of the calibrated points are different.

[0019] Preferably, the edge point calibration by traversal method further includes calibrating the qualified points and assigning channel values to the RGB values of the calibrated points: b ij = r ij = 0, where b ij represents the B channel value of the point (i, j), and r ij represents the R channel value of the point (i, j).

[0020] Preferably, in step 3, the insulator umbrella shape feature is used for recognition.

[0021] Preferably, in step 3, the contour image with edge points calibrated is traversed in two-dimensional pixels, so as to obtain the upper boundary, lower boundary, left boundary and right boundary of the insulator, and complete the construction of the insulator area, and then complete the recognition of the insulator.

[0022] Preferably, the two-dimensional pixel traversal is as follows:

[0023]

[0024] The number of insulator boundary points in each row is counted by the above formula and the number of insulator boundary points in each column where l cols is the number of rows of the contour image, and l rows is the number of columns of the contour image.

[0025] Preferably, the several insulators form an insulator string,

[0026] Obtaining the upper boundary of the insulator string:

[0027] Based on the statistically obtained data, the row threshold n1 is obtained, and the set of row numbers that meet the upper boundary condition is obtained according to the following formula:

[0028] N1 = {i|N(i)>i max - n1, N(i)<=i max}

[0029] Take the first row min(N1) in N1 as the upper boundary L of the insulator string FirstLine = min(N1);

[0030] Obtaining the lower boundary of the insulator string:

[0031] The set of row numbers that meet the lower boundary condition is obtained according to the following formula:

[0032] N2 = {i|N(i)>=i max - n1, n(i)<i max , n(i + 1)<i max - n1}

[0033] Take the last row in N2, max(N2), as the lower boundary L of the insulator string LastLine = max(N2);

[0034] Obtaining the left and right boundaries of the insulator string:

[0035] Count the number of boundary points in each column

[0036] Take the maximum value i among the number of column boundary points N(i) max = max(N(0), N(1),..., N(i - 1)), i = l row ;

[0037] Count the number of columns where the number of boundary points is equal to the maximum number of boundary points, and form a set of column numbers M = {j|f(b max ij , g max ij , r max ij ) = 1};

[0038] Take the minimum value min(M) and the maximum value max(M) in the set of column numbers M as the left boundary j left = min(M) and the right boundary j right = max(M), based on L FirstLine , L LastLine , j left , j right Construct the insulator region to complete the identification of the insulator string.

[0039] Preferably, step 4 includes: taking the upper and lower boundaries of the insulator as the range of longitudinal scanning, and counting the number of boundary points in each column within the scanning range, as shown in the following formula:

[0040]

[0041] Extract the columns with the characteristics of split columns. Since the number of boundary points D(j) of the split columns is much higher than the number of boundary points D(j - 1) and D(j + 1) of the surrounding columns, using the threshold d, according to the judgment condition of the following formula:

[0042] D location = {j|D(j) - D(j - 1) >= d, D(j) - D(j + 1) >= d},

[0043] Obtain the initial split queue D location , and at the same time count the distance between the two dividing lines, and take the interval distance with the most occurrences as the standard interval distance; starting from the leftmost side of the marked insulator string, using the standard interval as the split distance, obtain the standard split queue and split the boundary line of the insulator string into individual insulators.

[0044] Preferably, step 4 further includes adjusting the standard segmentation queue in combination with the initial segmentation queue:

[0045] Compare the points in the initial segmentation queue and the standard segmentation queue. For the same insulator segmentation position in both queues, obtain an initial segmentation queue point P1 and a standard segmentation queue point P2. When the distance between the two points is less than the threshold d, replace the point in the standard segmentation queue with the point in the initial segmentation queue; when the distance between the two points is greater than the threshold, ignore the point in the initial segmentation queue;

[0046] Use the adjusted standard segmentation queue and the boundary line of the insulator string to segment individual insulators.

[0047] The present invention has the following characteristics and beneficial effects:

[0048] Adopting the above technical solution, feature simplification is carried out using threshold binarization, and then the corrected infrared insulator image is obtained by image correction based on contour extraction. After obtaining the corrected infrared insulator image, the boundary points of the insulator are calibrated through one-dimensional pixel scanning, and then the distribution characteristics of the insulator boundary points are obtained through multi-dimensional scanning, realizing the extraction and segmentation of the insulator string. This method uses multi-dimensional scanning to extract the distribution characteristics of the insulator boundary points in the infrared image, and transfers the process of insulator string extraction and segmentation to the extraction of the distribution characteristics of the insulator boundary points, making the identification and segmentation of insulators in the infrared image have better portability, ease of use and easy function expansion. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0050] Figure 1 It is the principle flowchart of the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0052] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by terms such as "center", "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention. In addition, terms such as "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", etc. may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise stated, the meaning of "a plurality" is two or more.

[0053] In the description of the present invention, it should be noted that unless otherwise clearly specified and limited, the terms "mounted", "connected", "connected to" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood through specific situations.

[0054] The present invention provides a method for identifying and segmenting infrared images of insulators based on edge features, as Figure 1 shown, including the following steps:

[0055] Step 1: Collect images and make an original data set;

[0056] Step 2: Preprocess the images;

[0057] Step 2.1: Denoise the original data set;

[0058] Step 2.2: Perform threshold binaryzation using image features;

[0059] Step 2.3: Correct the image based on contour extraction;

[0060] Step 2.4: Calibrate edge points by traversing the corrected image;

[0061] Step 3: Identify according to insulator features;

[0062] Step 4: Extract image features to obtain the precise positioning and segmentation of a single insulator.

[0063] In the above technical solution, threshold binarization is used for feature simplification, and then image correction based on contour extraction is used to obtain the corrected infrared insulator image, laying a foundation for subsequent positioning and segmentation operations, reducing the amount of calculation and shortening the time consumption.

[0064] In addition, in step 1, an image is collected by an infrared camera, and the resolution of the infrared image is 640×480. It can be understood that the infrared image is convenient for correction. And through one-dimensional pixel scanning, the boundary points of the insulator are calibrated, and then through multi-dimensional scanning, the distribution characteristics of the insulator boundary points are obtained, realizing the extraction and segmentation of the insulator string. This method uses multi-dimensional scanning to extract the distribution characteristics of the insulator boundary points in the infrared image, transferring the process of insulator string extraction and segmentation to the extraction of insulator boundary point distribution characteristics, making the recognition and segmentation of the insulator in the infrared image not only reduce the amount of calculation and shorten the time consumption, but also have better portability, ease of use and are easy to expand functions, and the accuracy is greatly improved.

[0065] It can be understood that in step 2.1, the original data set is denoised by Gaussian filtering, thus laying a foundation for the subsequent accuracy. In addition, in step 2.3, the extracted contour image is corrected by one-dimensional pixel traversal.

[0066] In a further setting of the present invention, the edge point calibration by the traversal method includes calibrating the corrected contour image, and the calibrated points are the points within the insulator region on the image, and the previous pixel point and the next pixel point of the calibrated point are different. The edge point calibration by the traversal method also includes calibrating the qualified points and assigning channel values to the RGB values of the calibrated points: b ij =r ij =0, where b ij represents the B channel value of the point (i,j), and r ij represents the R channel value of the point (i,j).

[0067] In the above technical solution, the calibration of the edge points reduces the calculation amount of the subsequent vertical edge point quantity statistics.

[0068] In a further setting of the present invention, in step 3, identification is performed according to the umbrella shape feature of the insulator.

[0069] It can be understood that the umbrella shape feature is an inherent feature of the insulator

[0070] Specifically, in step 3, two-dimensional pixel traversal is performed on the contour image after the edge point calibration, and then the upper boundary, lower boundary, left boundary and right boundary of the insulator are obtained, and the construction of the insulator region is completed, and then the identification of the insulator is completed.

[0071] In the above technical solution, according to the umbrella-shaped inherent characteristics of the insulator, its contour is calibrated, and then the upper boundary, lower boundary, left boundary and right boundary of the insulator are obtained, reducing the calculation amount of the longitudinal edge point number statistics and greatly shortening the time consumption.

[0072] Further, the two-dimensional pixel traversal is as follows:

[0073]

[0074] The number of insulator boundary points in each row is counted by the above formula and the number of insulator boundary points in each column where l cols is the number of rows of the contour image, and l rows is the number of columns of the contour image.

[0075] In addition, the several insulators form an insulator string.

[0076] The upper boundary of the insulator string is obtained as follows:

[0077] Based on the statistically obtained data, the row threshold n1 is obtained, and the set of row numbers that meet the upper boundary condition is obtained according to the following formula:

[0078] N1 = {i|N(i) > i max - n1, N(i) <= i max}

[0079] Take the first row min(N1) in N1 as the upper boundary L of the insulator string FirstLine = min(N1);

[0080] The lower boundary of the insulator string is obtained as follows:

[0081] The set of row numbers that meet the lower boundary condition is obtained according to the following formula:

[0082] N2 = {i|N(i) >= i max - n1, n(i) < i max , n(i + 1) < i max - n1}

[0083] Take the last row max(N2) in N2 as the lower boundary L of the insulator string LastLine = max(N2);

[0084] The left boundary and right boundary of the insulator string are obtained as follows:

[0085] Count the number of boundary points in each column

[0086] Take the maximum value i of the column boundary point number N(i) max= max(N(0), N(1),..., N(i - 1)), i = l row ;

[0087] Count the number of columns where the number of boundary points is equal to the maximum number of boundary points, and form a set of column numbers M = {j|f(b max ij , g max ij , r max ij ) = 1};

[0088] Take the minimum value min(M) and the maximum value max(M) in the set of column numbers M as the left boundary j left = min(M) and the right boundary j right = max(M), and based on L FirstLine 、L LastLine 、j left 、j right Construct the insulator region, complete the identification of the insulator string, and narrow the traversal range of subsequent insulator separation, further reducing the computational amount.

[0089] A further setting of the present invention, the step 4 includes: using the upper and lower boundaries of the insulator as the range of longitudinal scanning, and counting the number of boundary points of each column within the scanning range, as shown in the following formula:

[0090]

[0091] Extract the columns with the characteristics of split columns. Since the number of boundary points D(j) of the split columns is much higher than the number of boundary points D(j - 1) and D(j + 1) of the surrounding columns, using the threshold d, according to the judgment condition of the following formula:

[0092] D location = {j|D(j) - D(j - 1) >= d, D(j) - D(j + 1) >= d},

[0093] Obtain the initial split queue D location , and at the same time count the distance between the two dividing lines, and take the interval distance with the most occurrences as the standard interval distance; starting from the leftmost side of the marked insulator string, using the standard interval as the split distance, obtain the standard split queue and split the boundary line of the insulator string into individual insulators.

[0094] Further, the step 4 further includes adjusting the standard split queue in combination with the initial split queue:

[0095] Compare the points in the initial segmentation queue and the standard segmentation queue. For the same insulator segmentation position in both queues, obtain a point P1 in the initial segmentation queue and a point P2 in the standard segmentation queue. When the distance between the two points is less than the threshold d, replace the point in the standard segmentation queue with the point in the initial segmentation queue; when the distance between the two points is greater than the threshold, ignore the point in the initial segmentation queue.

[0096] Use the adjusted standard segmentation queue and the boundary line of the insulator string to segment individual insulators.

[0097] It can be understood that in the above technical solution, by adjusting the initial separation queue using the standard interval, the interference of background objects is reduced, thereby improving the accuracy of recognition and segmentation.

[0098] Through the description of the above embodiments, those skilled in the art can understand that the technical solution of the embodiments of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. The software product is stored in a storage medium, including several instructions for causing a device (which can be a single-chip microcomputer, a chip, etc.) or a processor to execute all or part of the steps of the methods of the various embodiments of the present application. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs, etc., which can store program codes.

[0099] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, without departing from the principles and spirit of the present invention, various changes, modifications, substitutions, and variations to these embodiments, including components, still fall within the protection scope of the present invention.

Claims

1. An identification and segmentation method for infrared images of insulators based on edge features, characterized in that, It includes the following steps: Step 1: Collect images and create an original dataset; Step 2: Preprocess the images; Step 2.1: Denoise the original dataset; Step 2.2: Perform threshold binarization using image features; Step 2.3: Rectify the image based on contour extraction; Step 2.4: Calibrate the edge points by traversal method for the rectified image; Step 3: Identify according to insulator features; A number of the insulators form an insulator string, Obtaining the upper boundary of the insulator string: Based on the statistically obtained data, obtain the row threshold n1, and obtain the set of rows that meet the upper boundary condition according to the following formula: N1 = {i | N(i) > i max -n1, N(i) <= i max} Take the first row min(N1) in N1 as the upper boundary L of the insulator string FirstLine = min(N1); Obtaining the lower boundary of the insulator string: Obtain the set of rows that meet the lower boundary condition according to the following formula: N2 = {i | N(i) >= i max -n1, N(i) < i max , N(i + 1) < i max -n1} Take the last row in N2, max(N2), as the lower boundary L of the insulator string LastLine = max(N2); Obtaining the left and right boundaries of the insulator string: Count the number of boundary points in each column Take the maximum value i among the number of column boundary points N(i) max = max(N(0), N(1),..., N(i - 1)), i = l row ; Count the number of columns where the number of boundary points is equal to the maximum number of boundary points, and form a set of column numbers \(M = \{j|f(b maxij ,g maxij ,r maxij ) = 1\}\); Take the minimum value min(M) and the maximum value max(M) in the set M of column numbers as the left boundary j of the insulator string left = min(M) and the right boundary j right = max(M), based on L FirstLine 、L LastLine 、j left 、j right Construct the insulator region to complete the identification of the insulator string; Step 4: Extract image features to obtain precise positioning and segmentation of a single insulator, Taking the upper and lower boundaries of the insulator as the longitudinal scanning range, count the number of boundary points in each column within the scanning range, as shown in the following formula: Extract the columns with the characteristics of segmented columns. Since the number of boundary points D(j) of the segmented columns is much higher than the number of boundary points D(j - 1) and D(j + 1) of the surrounding columns, use the threshold d, according to the judgment condition of the following formula: D location = {j | D(j) - D(j - 1) >= d, D(j) - D(j + 1) >= d}, Obtain the initial segmentation queue D location , and at the same time, count the distances between the dividing lines of two adjacent insulators. Take the interval distance with the most occurrences as the standard interval distance; starting from the leftmost side of the marked insulator string, use the standard interval as the segmentation distance to obtain the standard segmentation queue, and segment individual insulators along the boundary line of the insulator string; Adjust the standard segmentation queue in combination with the initial segmentation queue: Compare the points in the initial segmentation queue and the standard segmentation queue. For the same insulator segmentation position in the two queues, obtain an initial segmentation queue point P1 and a standard segmentation queue point P2. When the distance between the two points is less than the threshold d, replace the point in the standard segmentation queue with the point in the initial segmentation queue; when the distance between the two points is greater than the threshold, ignore the point in the initial segmentation queue; Use the adjusted standard segmentation queue and the boundary line of the insulator string to segment a single insulator.

2. The identification and segmentation method of the insulator infrared image based on edge features according to claim 1, characterized in that In Step 1, images are collected by an infrared camera. In Step 2.1, the original dataset is denoised by Gaussian filtering. In Step 2.3, the extracted contour image is rectified by one-dimensional pixel traversal.

3. The method for identifying and segmenting the infrared image of an insulator based on edge features according to claim 1, wherein, The traversal method for edge point calibration includes calibrating the rectified contour image. The calibrated points are the points within the insulator area on the image, and the previous pixel point and the next pixel point of the calibrated points are different.

4. The method for identifying and segmenting the infrared image of an insulator based on edge features according to claim 3, characterized in that, The traversal method for edge point calibration further includes calibrating the qualified points and assigning channel values to the RGB values of the calibrated points: b ij = r ij = 0, where b ij represents the B-channel value of the point (i, j), and r ij represents the R-channel value of the point (i, j).

5. The identification and segmentation method of the insulator infrared image based on edge features according to claim 4, characterized in that, In Step 3, identification is performed according to the umbrella shape feature of the insulator.

6. The method for identifying and segmenting an infrared image of an insulator based on edge features according to claim 5, characterized in that In Step 3, two-dimensional pixel traversal is performed on the contour image after edge point calibration, so as to obtain the upper boundary, lower boundary, left boundary and right boundary of the insulator, and complete the construction of the insulator area, and then complete the identification of the insulator.

7. The method for identifying and segmenting an infrared image of an insulator based on edge features according to claim 6, characterized in that The two-dimensional pixel traversal is as follows: Count the number of insulator boundary points in each row using the above formula and the number of insulator boundary points in each column where l cols is the number of rows of the contour image, and l rows is the number of columns of the contour image.

Images