Zero value detection method and system for power grid based on wavelet transform picture fusion

Abstract

The invention provides a zero value detection method and system for a power grid based on wavelet transform picture fusion, and relates to the technical field of image classification, and the method comprises the steps: carrying out the wavelet decomposition of an infrared image and a visible light image, and obtaining low-frequency components and high-frequency components in horizontal, vertical and diagonal directions, the low-frequency component is composed of a first low-frequency component and a second low-frequency component, and the high-frequency component is composed of a first high-frequency component and a second high-frequency component; fusing the first low-frequency component and the second low-frequency component based on a weighted average fusion rule to obtain a first fusion component, and fusing the first high-frequency component and the second high-frequency component based on a selection rule to obtain a second fusion component; and reconstructing the first fusion component and the second fusion component based on inverse wavelet transform to obtain a final fusion image. According to the invention, zero value detection can be automatically carried out through the infrared picture and the visible light picture shot by the unmanned aerial vehicle without manual inspection, the temperature reading error is reduced, and the picture zero value detection efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image classification, and in particular to a zero value detection method and system for a power grid based on wavelet transform picture fusion. Background Art

[0002] As an important technical means to replace manual inspections, autonomous drone inspections have been vigorously promoted by power grid companies. Intelligent defect recognition of drone inspection results is an important part of reducing labor costs. It can greatly improve inspection efficiency and reduce the workload of grassroots inspectors, and has important application scenarios.

[0003] However, after the drone airport patrol operation is completed, each infrared image of the inspection still needs to be manually analyzed. There is a high possibility of error in temperature reading by judging the color circle points with the naked eye, and the detection efficiency is low.

[0004] Therefore, it is urgent to provide a solution to improve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a zero value detection method and system for a power grid based on wavelet transform image fusion, so as to improve the problem that the temperature reading by visually judging the color circle point is more likely to produce errors and has low efficiency.

[0006] In the first aspect, the present invention provides a zero value detection method for a power grid based on wavelet transform image fusion, which adopts the following technical solution: Perform wavelet decomposition on the infrared image and the visible light image to obtain low-frequency components and high-frequency components in horizontal, vertical and diagonal directions, respectively, wherein the low-frequency components are composed of a first low-frequency component and a second low-frequency component, and the high-frequency components are composed of a first high-frequency component and a second high-frequency component; fusing the first low-frequency component and the second low-frequency component based on a weighted average fusion rule to obtain a first fused component, and fusing the first high-frequency component and the second high-frequency component based on a selection rule to obtain a second fused component; The first fused component and the second fused component are reconstructed based on inverse wavelet transform to obtain a final fused image.

[0007] The beneficial effect of the zero value detection method for a power grid based on wavelet transform image fusion provided by the present invention is that the present invention can automatically perform zero value detection through infrared images and visible light images taken by a drone without manual inspection, thereby reducing the error of temperature reading and improving the efficiency of image zero value detection.

[0008] Optionally, the high-frequency components include high-frequency components in horizontal, vertical and diagonal directions, which represent edges and details of the image, and the low-frequency components represent general contour information of the image.

[0009] Optionally, the wavelet decomposition is implemented according to a wavelet basis function, wherein: The mathematical expression of the number of decomposition layers of the wavelet decomposition is: ; in, represents the number of decomposition levels, Indicates the width of the image. Indicates the height of the image.

[0010] The mathematical expression of the wavelet decomposition is: ; in, represents the low-frequency component, Represents the high-frequency component in the horizontal direction, Represents the high-frequency component in the vertical direction, represents the high-frequency component in the diagonal direction, represents the wavelet decomposition function.

[0011] Optionally, the first fusion component is obtained based on weighted average, and the mathematical expression is: ; ; in, represents the first fusion component, represents the low-frequency weight of infrared light, represents the low-frequency weight of visible light, Represents the low-frequency component after infrared light decomposition, Represents the low-frequency component after the decomposition of the visible light image, represents the entropy value of the infrared image, Represents the entropy value of the visible light image; The second fusion component is obtained based on a selection rule, including the fusion of high-frequency components of the picture in the horizontal direction, the vertical direction and the diagonal direction, and the mathematical expression is: ; in, represents the second fusion component, Represents the high-frequency components of infrared images in the horizontal, vertical and diagonal directions. Represents the high-frequency components of the visible light image in the horizontal, vertical and diagonal directions.

[0012] Optionally, the mathematical expression of the final fused image is: ; in, represents the final fused image, represents the inverse wavelet transform operation, represents the first fusion component, Represents the low-frequency component after infrared light decomposition, Represents the low-frequency component after the decomposition of the visible light image, Represents the high-frequency component in the horizontal direction after fusion, Represents the high-frequency component in the vertical direction after fusion, Represents the high-frequency component in the diagonal direction after fusion, represents the low-frequency weight of infrared light, Represents the low-frequency weight of visible light.

[0013] In a second aspect, the present invention also provides a zero value detection system for a power grid based on wavelet transform image fusion, comprising: a data input module, a defect analysis module and a defect result output module, wherein: The data input module is used to obtain and input image data, wherein the image data consists of visible light image data and infrared image data, wherein the image data can be accessed by online backtransmission from the UAV flight control end, or by offline uploading of images; The defect analysis module is used to generate a fused image with a temperature value function and an insulator outline after performing wavelet transformation on the image data input by the data input module; The defect result output module is used to output the defect result after marking the highest temperature and the lowest temperature at the defect position of the fused image according to the standard defect library.

[0014] Optionally, the defect analysis module specifically includes an infrared image information reading component, a visible light infrared image fusion component, a fused image segmentation component and a defect analysis component, wherein: The infrared image information reading component is used to pre-process the infrared temperature data before image fusion to obtain the mapping relationship between the grayscale value of the infrared image pixel before fusion and the temperature value function; The visible light infrared image fusion component is used to fuse the visible light and infrared images according to the wavelet transform, and retain the temperature characteristics of the infrared image during the fusion process, fuse the low-frequency components based on the weighted average fusion rule, and fuse the high-frequency components based on the selection fusion rule; The fused image segmentation component is used to segment the image fused by the visible light image fusion component, and obtain the outline of the insulator in the fused image as the input of the defect analysis component; The defect analysis component is used to extract the infrared grayscale value of the insulator contour image, and obtain the insulator contour temperature value function and the corresponding maximum temperature, minimum temperature, and temperature difference based on the mapping relationship between the grayscale value and the temperature value function obtained by the infrared image information reading component.

[0015] Optionally, the mathematical expression of the temperature value function is: ; in, is the temperature value function, is the proportionality factor, determined by the infrared device calibration, is the offset, is the gray value.

[0016] For the beneficial effects of the second aspect, please refer to the description of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flowchart showing a method for detecting zero value of a power grid based on infrared images provided by the present invention; Figure 2 A schematic diagram showing the overall architecture of a power grid system based on infrared image zero value detection provided by the present invention; Figure 3 A schematic diagram showing the design of modules and functional components of a power grid system based on infrared image zero value detection provided by the present invention; Figure 4 A schematic diagram showing an algorithm of a system for zero-value detection of a power grid based on infrared images provided by the present invention; Figure 5 A schematic diagram of image fusion of a power grid system based on infrared image zero value detection provided by the present invention is shown. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Unless otherwise defined, the technical terms or scientific terms used herein should be understood by people with general skills in the field to which the present invention belongs. "Including" and similar words used in this article mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects.

[0019] The embodiment of the present invention provides a zero value detection method for a power grid based on wavelet transform image fusion, see Figure 1 , including the following steps: S1. Perform wavelet decomposition on the infrared image and the visible light image to obtain low-frequency components and high-frequency components in the horizontal, vertical and diagonal directions, respectively, wherein the low-frequency components are composed of a first low-frequency component and a second low-frequency component, and the high-frequency components are composed of a first high-frequency component and a second high-frequency component; S2. Based on a weighted average fusion rule, the first low-frequency component and the second low-frequency component are fused to obtain a first fused component, and based on a selection rule, the first high-frequency component and the second high-frequency component are fused to obtain a second fused component; S3. Reconstruct the first fusion component and the second fusion component based on inverse wavelet transform to obtain a final fusion image.

[0020] In some embodiments, when executing step S1, the wavelet decomposition is implemented according to the wavelet basis function, and the process is: the input infrared image and visible light images The wavelet coefficients are decomposed into multi-scale and multi-resolution wavelet coefficients, wherein the wavelet coefficients include low-frequency components and high-frequency components.

[0021] Specifically, the high-frequency components include high-frequency components in horizontal, vertical and diagonal directions, which represent edges and details of the image, and the low-frequency components represent general contour information of the image.

[0022] In fact, choosing a suitable wavelet basis function can take into account both smoothness and detail information. In order to prevent the number of decomposition layers from being too small to capture detail information or too many decomposition layers from causing excessive sparseness of information and affecting computational efficiency, the present invention selects a suitable number of decomposition layers according to the resolution and features of the image. Perform wavelet decomposition of the image, where The mathematical expression of the number of decomposition layers of the wavelet decomposition is: ; in, represents the number of decomposition levels, Indicates the width of the image. Indicates the height of the image.

[0023] The mathematical expression of the wavelet decomposition is: ; in, represents the low-frequency component, Represents the high-frequency component in the horizontal direction, Represents the high-frequency component in the vertical direction, represents the high-frequency component in the diagonal direction, represents the wavelet decomposition function.

[0024] Furthermore, for infrared images and visible light images By performing discrete wavelet transform, we can get: ; ; in, represents the discrete wavelet transform operation, Represents the low-frequency component after infrared light decomposition, Represents the high-frequency component in the horizontal direction after the infrared image is decomposed. Represents the high-frequency component in the vertical direction after the infrared image is decomposed. Represents the high-frequency component in the diagonal direction after the infrared image is decomposed. represents the input infrared image, Represents the low-frequency component after the decomposition of the visible light image, Represents the high-frequency component in the horizontal direction after the visible light image is decomposed. Represents the high-frequency component in the vertical direction after the visible light image is decomposed. Represents the high-frequency component in the diagonal direction after the visible light image is decomposed. Indicates the discrete wavelet transform operation on the visible light image. Represents the input visible light image.

[0025] In some embodiments, when executing step S2, the first fusion component is obtained based on weighted average, and the mathematical expression is: ; ; in, represents the first fusion component, represents the low-frequency weight of infrared light, represents the low-frequency weight of visible light, Represents the low-frequency component after infrared light decomposition, Represents the low-frequency component after the decomposition of the visible light image, represents the entropy value of the infrared image, Represents the entropy value of the visible light image.

[0026] The second fusion component is obtained based on a selection rule, including the fusion of high-frequency components of the picture in the horizontal direction, the vertical direction and the diagonal direction, and the mathematical expression is: ; in, represents the second fusion component, Represents the high-frequency components of infrared images in the horizontal, vertical and diagonal directions. Represents the high-frequency components of the visible light image in the horizontal, vertical and diagonal directions.

[0027] In some embodiments, when executing step S2, the mathematical expression of the final fused image is: ; in, represents the final fused image, represents the inverse wavelet transform operation, represents the first fusion component, Represents the low-frequency component after infrared light decomposition, Represents the low-frequency component after the decomposition of the visible light image, Represents the high-frequency component in the horizontal direction after fusion, Represents the high-frequency component in the vertical direction after fusion, Represents the high-frequency component in the diagonal direction after fusion, represents the low-frequency weight of infrared light, Represents the low-frequency weight of visible light.

[0028] The embodiment of the present invention further provides a zero-value detection system for a power grid based on wavelet transform image fusion, comprising: a data input module, a defect analysis module and a defect result output module.

[0029] The data input module is used to obtain and input image data, and the image data consists of visible light image data and infrared image data. The image data can be accessed through online transmission from the UAV flight control end, or through offline uploading of images.

[0030] Specifically, the source of the image data includes but is not limited to images of drone inspections, images of manual inspections, and images extracted from other external systems, and the formats of the images include but are not limited to JPG, PNG, BMP and other formats.

[0031] In fact, the input format of the data input module is a visible light picture or an infrared picture, which are distinguished by picture annotation, and the pictures are classified and stored as a data source for use by the defect analysis module.

[0032] The defect analysis module is used to generate a fused image with a temperature value function and an insulator outline after performing wavelet transformation on the image data input by the data input module.

[0033] Specifically, the defect analysis module is the core module of the zero-value detection system. Its main function is to perform wavelet transformation fusion on the visible light image input by the data input module and the infrared image to obtain a fused image. The fused image has the temperature value function of the infrared image and the characteristics of visible light. Compared with directly using infrared images for segmentation, the segmentation accuracy of the fused image is significantly improved. DeeplabV3+ is used to segment the fused image to obtain the position of the insulator in the image. For the segmented image, the original infrared component is used to extract the infrared temperature value function to obtain the highest and lowest temperatures of the insulator part, as well as the corresponding temperature difference.

[0034] The defect result output module is used to output the defect result after marking the highest temperature and the lowest temperature at the defect position of the fused image according to the standard defect library.

[0035] Specifically, the specific process of outputting defect results is as follows: Determine whether it exceeds the threshold. If it exceeds the threshold, the defect output in the mapping standard library is marked, and the image is marked as a defect image, marking the highest temperature, lowest temperature and temperature difference at the defect position of the image.

[0036] Furthermore, the defect analysis module specifically includes an infrared image information reading component, a visible light infrared image fusion component, a fused image segmentation component and a defect analysis component, wherein: The infrared image information reading component is used to pre-process the infrared temperature data before the image fusion to obtain the grayscale value of the infrared image pixel before fusion Function of temperature The mapping relationship of The visible light infrared image fusion component is used to fuse the visible light and infrared images according to the wavelet transform, and retain the temperature characteristics of the infrared image during the fusion process, fuse the low-frequency components based on the weighted average fusion rule, and fuse the high-frequency components based on the selection fusion rule; The fused image segmentation component is used to segment the image fused by the visible light image fusion component, and obtain the outline of the insulator in the fused image as the input of the defect analysis component.

[0037] Specifically, the fused image segmentation component is based on the DeeplabV3+ model. DeeplabV3+ is a semantic segmentation model based on a convolutional neural network. It achieves high-precision segmentation by extracting image features and generating category probabilities for each pixel. The image after model segmentation is a mask, and each pixel value represents its category.

[0038] ; in, is the segmentation mask, at the pixel position The value is the category .

[0039] Furthermore, the image processing tool OpenCV is used to segment the boundaries of the region from the mask and draw them on the image.

[0040] The defect analysis component is used to extract the infrared grayscale value of the insulator contour image, and obtain the insulator contour temperature value function and the corresponding maximum temperature, minimum temperature, and temperature difference based on the mapping relationship between the grayscale value and the temperature value function obtained by the infrared image information reading component.

[0041] Specifically, the defect analysis component records from the fusion process and Extract the infrared low-frequency and high-frequency components of the pixel, combine the low-frequency and high-frequency components, and reconstruct the infrared grayscale value of the pixel: ); According to the infrared gray value , we can get the temperature value function corresponding to the pixel point, and then get the highest temperature, lowest temperature, and temperature difference of the insulator outline in the image. The mathematical expression of the temperature value function is: ; in, is the temperature value function, is the proportionality factor, determined by the infrared device calibration, is the offset, is the gray value.

[0042] See also Figure 2 The present invention provides a zero value detection system for a power grid based on wavelet transform image fusion, which includes: a data input module 101, a defect analysis module 102 and a defect result output module 103.

[0043] See also Figure 3 The data input module inputs visible light images and infrared images respectively. After receiving them, the defect analysis module reads the infrared image information and fuses the visible light infrared images, and performs fused image segmentation and defect analysis. Finally, the defect result output module maps the standard defect library output and outputs the defect result after image marking.

[0044] See also Figure 4 , including: input infrared and visible light images; read infrared temperature information; fuse based on wavelet transform; segment the fused image to identify the insulator outline; read the temperature information within the insulator outline of the fused image; obtain the highest and lowest temperatures of the insulator; obtain the temperature difference parameter n; map the standard library defect output, and output after the image is marked.

[0045] Specifically, after obtaining the temperature difference parameter n, it is necessary to determine whether the temperature difference of the same insulator exceeds the threshold according to the temperature threshold. If so, the standard library defect output is mapped and the image is marked and output. If not, the process ends.

[0046] See also Figure 5 , including three processes: image wavelet decomposition S401, component fusion S402, and wavelet reconstruction S403.

[0047] Specifically, first, the infrared image is decomposed by wavelet to obtain the high-frequency coefficients and low-frequency coefficients of the infrared image, and the visible light image is decomposed by wavelet to obtain the high-frequency coefficients and low-frequency coefficients of the visible light image. Secondly, the high-frequency coefficients of the infrared image and the high-frequency coefficients of the visible light image are fused based on a selection rule fusion method, and the low-frequency coefficients of the infrared image and the low-frequency coefficients of the visible light image are fused based on a weighted average fusion method. Finally, the fused image is obtained through wavelet reconstruction.

[0048] Although the embodiments of the present invention are described in detail above, it is obvious to those skilled in the art that various modifications and variations can be made to these embodiments. However, it should be understood that such modifications and variations are within the scope and spirit of the present invention as described in the claims. Moreover, the present invention described herein may have other embodiments and may be implemented or realized in a variety of ways.

Claims

1. A zero value detection method for power grid based on wavelet transform image fusion, characterized in that: include: Perform wavelet decomposition on the infrared image and the visible light image to obtain low-frequency components and high-frequency components in horizontal, vertical and diagonal directions, respectively, wherein the low-frequency components are composed of a first low-frequency component and a second low-frequency component, and the high-frequency components are composed of a first high-frequency component and a second high-frequency component; fusing the first low-frequency component and the second low-frequency component based on a weighted average fusion rule to obtain a first fused component, and fusing the first high-frequency component and the second high-frequency component based on a selection rule to obtain a second fused component; The first fused component and the second fused component are reconstructed based on inverse wavelet transform to obtain a final fused image.

2. A zero value detection method based on wavelet transform image fusion in power grid according to claim 1, characterized in that: The high-frequency components include high-frequency components in horizontal, vertical and diagonal directions, which represent the edges and details of the image, and the low-frequency components represent the general contour information of the image.

3. The zero value detection method based on wavelet transform image fusion in power grid according to claim 1 is characterized in that: The wavelet decomposition is implemented according to wavelet basis functions, wherein: The mathematical expression of the number of decomposition layers of the wavelet decomposition is: ； in, represents the number of decomposition levels, Indicates the width of the image. Indicates the height of the image; The mathematical expression of the wavelet decomposition is: ； in, represents the low-frequency component, Represents the high-frequency component in the horizontal direction, Represents the high-frequency component in the vertical direction, represents the high-frequency component in the diagonal direction, represents the wavelet decomposition function.

4. The zero value detection method based on wavelet transform image fusion in power grid according to claim 1 is characterized in that: The first fusion component is obtained based on weighted average, and the mathematical expression is: ； ； in, represents the first fusion component, represents the low-frequency weight of infrared light, represents the low-frequency weight of visible light, Represents the low-frequency component after infrared light decomposition, Represents the low-frequency component after the decomposition of the visible light image, represents the entropy value of the infrared image, Represents the entropy value of the visible light image; The second fusion component is obtained based on a selection rule, including the fusion of high-frequency components of the picture in the horizontal direction, the vertical direction and the diagonal direction, and the mathematical expression is: ； in, represents the second fusion component, Represents the high-frequency components of infrared images in the horizontal, vertical and diagonal directions. Represents the high-frequency components of the visible light image in the horizontal, vertical and diagonal directions.

5. The zero value detection method based on wavelet transform image fusion in power grid according to claim 1 is characterized in that: The mathematical expression of the final fused image is: ； in, represents the final fused image, represents the inverse wavelet transform operation, represents the first fusion component, Represents the low-frequency component after infrared light decomposition, Represents the low-frequency component after the decomposition of the visible light image, Represents the high-frequency component in the horizontal direction after fusion, Represents the high-frequency component in the vertical direction after fusion, Represents the high-frequency component in the diagonal direction after fusion, represents the low-frequency weight of infrared light, Represents the low-frequency weight of visible light.

6. A zero value detection system for power grid based on wavelet transform image fusion, characterized in that: include: Data input module, defect analysis module and defect result output module, among which, The data input module is used to obtain and input image data, wherein the image data consists of visible light image data and infrared image data, wherein the image data can be accessed by online backtransmission from the UAV flight control end, or by offline uploading of images; The defect analysis module is used to generate a fused image with a temperature value function and an insulator outline after performing wavelet transformation on the image data input by the data input module; The defect result output module is used to output the defect result after marking the highest temperature and the lowest temperature at the defect position of the fused image according to the standard defect library.

7. A zero value detection system based on wavelet transform image fusion in power grid according to claim 6, characterized in that: The defect analysis module specifically includes an infrared image information reading component, a visible light infrared image fusion component, a fused image segmentation component and a defect analysis component, wherein: The infrared image information reading component is used to pre-process the infrared temperature data before image fusion to obtain the mapping relationship between the grayscale value of the infrared image pixel before fusion and the temperature value function; The visible light infrared image fusion component is used to fuse the visible light and infrared images according to the wavelet transform, and retain the temperature characteristics of the infrared image during the fusion process, fuse the low-frequency components based on the weighted average fusion rule, and fuse the high-frequency components based on the selection fusion rule; The fused image segmentation component is used to segment the image fused by the visible light image fusion component, and obtain the outline of the insulator in the fused image as the input of the defect analysis component; The defect analysis component is used to extract the infrared grayscale value of the insulator contour image, and obtain the insulator contour temperature value function and the corresponding maximum temperature, minimum temperature, and temperature difference based on the mapping relationship between the grayscale value and the temperature value function obtained by the infrared image information reading component.

8. A zero value detection system based on wavelet transform image fusion in power grid according to claim 7, characterized in that: The mathematical expression of the temperature value function is: ； in, is the temperature value function, is the proportionality factor, determined by the infrared device calibration, is the offset, is the gray value.

Images