Tower inclination detection method based on high-resolution SAR image

Abstract

The invention discloses a tower inclination detection method based on a high-resolution SAR image, and relates to the technical field of tower inclination monitoring, and the method comprises the steps: SAR data set manufacturing: cutting the high-resolution SAR image, manufacturing a data set, and enabling each image to comprise a plurality of tower targets; according to the method, image threshold segmentation is carried out on strong scattering point information of an SAR image, then image segmentation is carried out by using a morphological method, contour information of a tower is extracted, the inclination angle of the tower is obtained and calculated by extracting a tower contour circumscribed rectangle, and finally, the inclination angle of the tower is calculated according to the relative geometrical relationship among a plurality of towers in the same image. According to the method, the inclined pole tower is distinguished and extracted, the investment of manpower, material resources and financial resources is saved to a great extent, satellite inspection is used, a large-area pole tower can be covered through one-time shooting, excessive manual interference is not needed, the inspection efficiency is greatly improved, the investment of line operation and maintenance personnel is reduced, and the method has the advantages of short period, wide inspection area and the like.

Description

Technical Field

[0001] The invention relates to the technical field of pole tower tilt monitoring, and in particular to a pole tower tilt detection method based on high-resolution SAR images. Background Art

[0002] Transmission lines are the lifeline of power transmission. Transmission towers are prone to tilting and tower collapse due to factors such as floods and external force damage, causing huge economic losses and safety hazards. Therefore, transmission tower tilt monitoring is of great significance to ensure the safe operation of the power grid. Current tower tilt monitoring methods are mostly based on manual inspections, communication technology, and aerial images from drones and helicopters.

[0003] At present, manual inspections require manual field surveys, which are greatly affected by the terrain and are inefficient. The initial installation and subsequent maintenance costs of wired communication technology are high, and wireless communication requires long-term mobile communication rental services with high operating costs. UAV inspections are costly and have poor real-time performance, making it impossible to monitor the status of transmission lines online. Therefore, a tower tilt detection method based on high-resolution SAR images is urgently needed. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that the current vehicle drivers and passengers control the opening and closing of doors by means of switch buttons, wherein the process is as follows: the driver and passengers press the button, the button sends a signal, and the controller receives the signal to control the opening and closing of the door. This process is completely operated manually, resulting in a delay in the opening and closing of the door, and the vehicle cannot predict in advance and prepare to open the door in advance to save the waiting time for opening and closing the door. A tower tilt detection method based on high-resolution SAR images is provided.

[0005] To achieve the above object, the present invention provides the following technical solution: a tower tilt detection method based on high-resolution SAR images, comprising: SAR data set production: crop high-resolution SAR images to produce data sets, with each image containing multiple tower targets; Image preprocessing and binary segmentation: perform lee filtering on the image, use Otsu global threshold algorithm for binary extraction, classify the points with pixel grayscale values ​​greater than the threshold as the target, marked as 255 grayscale, and the rest as the background, marked as 0 grayscale, to complete the image segmentation; Morphological processing: The binary image is morphologically processed by using morphological opening operation, which first corrodes the image and then expands it to eliminate the influence of small noise points on the extraction of the tower, while retaining the shape characteristics of the tower as much as possible; Enclosing rectangle extraction: extract the enclosing rectangle of the image after morphological processing, and calculate the minimum enclosing rectangle based on the geometric features of the contour; Obtaining the inclination angle of the tower: construct the inclination angle of the tower through the angle between the longitudinal side of the circumscribed rectangle and the vertical direction; Determine the geometric relationship between poles and towers: mainly involving the height of the towers, the distance between the lines and the towers, the horizontal and vertical distances between the lines, and the arrangement of lightning conductors on the towers; Tower tilt state identification: By extracting the tilt angle of the tower, when the difference between the tilt angle of a tower and the tilt angle of other towers in the image is greater than a certain threshold, it is defined as a tilted tower.

[0006] Preferably, the key technologies for high-resolution SAR image processing include motion compensation technology, imaging algorithm, image enhancement technology, and target recognition and extraction technology.

[0007] Preferably, the lee filter is based on the window function principle, estimates the noise level by calculating the mean and variance of the pixel values ​​in the neighborhood around each pixel point, and performs filtering processing. The steps of the Otsu global threshold algorithm are: preprocessing, grayscale histogram, threshold calculation, and binarization.

[0008] Preferably, in the morphological processing, the opening operation is an operation in the image morphological processing, which consists of two steps: first corrosion and then expansion. The purpose of the opening operation is to eliminate small noise points in the image while retaining the structural features of large objects in the image.

[0009] Preferably, in the extraction of the bounding rectangle, the bounding rectangle in the SAR image is extracted by using the findContours method in the OpenCV library, which includes the following steps: edge detection, contour detection, and bounding rectangle extraction.

[0010] Preferably, in the extraction of the bounding rectangle, the center point, width, height and rotation angle of the minimum bounding rectangle of the contour are first calculated, and then this information is encapsulated in a Box2D object and returned. This minimum bounding rectangle can be used to describe the overall shape and direction of the contour.

[0011] Preferably, in acquiring the inclination angle of the pole tower, based on the geometric analysis of the pole tower structure, the angle between the circumscribed rectangle side and the vertical direction is measured and calculated, thereby obtaining the specific inclination angle of the pole tower.

[0012] Preferably, in the determination of the tilt state of the tower, a threshold is set according to standard regulations, and the detected tower tilt angle is compared with other towers in the image. If the tilt angle difference exceeds the threshold, it is defined as the tower tilted; if the tilt angle difference is lower than the threshold, the tower is not tilted.

[0013] Compared with the prior art, the present invention has the following beneficial effects: In the present invention, image threshold segmentation is performed on the strong scattering point information of the SAR image, and then the image segmentation is performed using a morphological method to extract the contour information of the pole tower. The inclination angle of the pole tower is obtained and calculated by extracting the circumscribed rectangle of the pole tower contour. Finally, the inclined pole tower is identified and extracted through the relative geometric relationship between multiple pole towers in the same image, which saves the investment of manpower, material resources and financial resources to a great extent. Moreover, by using satellite inspection, a large area of ​​pole towers can be covered in one shot without too much manual intervention, which greatly improves the inspection efficiency and reduces the investment of line operation and maintenance personnel. It has the advantages of short cycle and wide inspection area, and is of great significance to the maintenance of transmission line protection and the safe operation of power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0016] See also Figure 1 , the tower tilt detection method based on high-resolution SAR images includes: SAR data set production: crop high-resolution SAR images to produce data sets, with each image containing multiple tower targets; Image preprocessing and binary segmentation: perform lee filtering on the image, use Otsu global threshold algorithm for binary extraction, classify the points with pixel grayscale values ​​greater than the threshold as the target, marked as 255 grayscale, and the rest as the background, marked as 0 grayscale, to complete the image segmentation; Morphological processing: The binary image is morphologically processed by using morphological opening operation, which first corrodes the image and then expands it to eliminate the influence of small noise points on the extraction of the tower, while retaining the shape characteristics of the tower as much as possible; Enclosing rectangle extraction: extract the enclosing rectangle of the image after morphological processing, and calculate the minimum enclosing rectangle based on the geometric features of the contour; Obtaining the inclination angle of the tower: construct the inclination angle of the tower through the angle between the longitudinal side of the circumscribed rectangle and the vertical direction; Determine the geometric relationship between poles and towers: mainly involving the height of the towers, the distance between the lines and the towers, the horizontal and vertical distances between the lines, and the arrangement of lightning conductors on the towers; Tower tilt state identification: By extracting the tilt angle of the tower, when the difference between the tilt angle of a tower and the tilt angle of other towers in the image is greater than a certain threshold, it is defined as a tilted tower.

[0017] Example 1 As a preferred embodiment of the present invention: key technologies for high-resolution SAR image processing include motion compensation technology, imaging algorithm, image enhancement technology, and target recognition and extraction technology; Motion compensation technology: During SAR imaging, the instability of platform motion will cause phase errors in echo signals, thus affecting image quality. Therefore, motion compensation technology is an indispensable part of SAR imaging processing. It eliminates phase errors and improves image quality by accurately estimating and compensating platform motion errors. Imaging algorithm: Imaging algorithm is the core technology of high-resolution SAR imaging processing. Its purpose is to convert echo signals into high-quality images. At present, commonly used imaging algorithms include range Doppler algorithm (RDA), polar format algorithm (PFA) and XR algorithm. Image enhancement technology: High-resolution SAR images usually have problems such as speckle noise and geometric distortion, and image enhancement processing is required. Image enhancement technologies include filtering, interpolation, super-resolution, etc., which can effectively improve image quality and improve image availability and readability. Target recognition and extraction technology: Target recognition and extraction of high-resolution SAR images is one of the important applications of SAR imaging processing. This technology realizes automatic recognition and extraction of targets in SAR images through image segmentation, feature extraction, classification recognition and other methods, providing support for subsequent decision-making and application.

[0018] Example 2 As a preferred embodiment of the present invention: Lee filtering is based on the principle of window function, and estimates the noise level by calculating the mean and variance of the pixel values ​​in the neighborhood around each pixel point, and performs filtering processing. The steps of the Otsu global threshold algorithm are: preprocessing, grayscale histogram, threshold calculation, and binarization; Step 1: Select a fixed-size neighborhood window. The window size is usually 3x3 or 5x5. For pixels on the edge of the window, you can choose to extend the edge or ignore it. Step 2: Traverse each pixel in the image, and perform the following steps for each pixel: extract pixel values ​​in the neighborhood window, and calculate the mean and variance of the pixel values ​​in the window; use the variance to estimate the noise variance of the current pixel. The common estimation method is to add the variances of all pixels in the window and divide it by the window size; calculate the estimated value of the signal-to-noise ratio (SNR), which is usually the signal mean divided by the square root of the noise variance; judge according to the pre-set signal-to-noise ratio threshold. If the signal-to-noise ratio is less than the threshold, it is considered that the current pixel is subject to more noise interference and a filtering operation is performed. Otherwise, the original pixel value is retained; for the pixel points that need to be filtered, use linear weighted smoothing to filter them. The formula for calculating the filtered pixel value is: filtered pixel value = original pixel value + weight * (window mean - original pixel value), where the weight can be adjusted according to actual needs, usually between 0 and 1. Lee filtering reduces the noise in the image and improves the image quality and readability; in the Otsu global threshold algorithm; The method of Otsu global threshold algorithm is: Step 1: Preprocessing: preprocess the input image to reduce noise and enhance image features. The common preprocessing method is Gaussian smoothing filtering, which helps smooth the image and reduce random noise. Step 2: Grayscale histogram, calculate the grayscale histogram of the image, that is, count the frequency of each grayscale level in the image. The histogram can provide the grayscale distribution information of the image; Step 3: Threshold calculation is the key step of the Otsu algorithm. The algorithm traverses all possible thresholds and calculates the ratio of the inter-class variance to the intra-class variance corresponding to each threshold. Otsu's goal is to find a value that maximizes this ratio. Minimizing the intra-class variance means that the pixels on both sides of the threshold are as similar as possible, while maximizing the inter-class variance means that the difference between the pixels on both sides of the threshold is as large as possible.

[0019] Step 4: Binarization: Use the calculated threshold T to binarize the original image. All pixels less than or equal to the threshold T are set as background pixels (e.g. 0), and all pixels greater than the threshold T are set as foreground pixels (e.g. 255).

[0020] Example 3 As a preferred embodiment of the present invention: in morphological processing, the opening operation is an operation in image morphological processing, which consists of two steps: first, corrosion, and then expansion. The purpose of the opening operation is to eliminate small noise points in the image while retaining the structural features of large objects in the image; Mathematical expression of opening operation: Given an input image I and a structural element B, the opening operation can be expressed by the mathematical expression: Opening(I,B)=Dilation(Erosion(I,B),B), where Erosion(I,B) represents the result of the input image I being eroded by the structural element B, and Dilation(·) represents the dilation operation.

[0021] Example 4 As a preferred embodiment of the present invention: in the extraction of the bounding rectangle, the bounding rectangle in the SAR image is extracted by the findContours method in the OpenCV library, which includes the following steps: edge detection, contour detection, and bounding rectangle extraction; Edge detection: First, the image needs to be edge detected, which is usually done by converting it to a grayscale image and applying threshold processing to separate the target object from the background; Contour detection: Use the findContours function to detect contours in the image. This function can detect contours in the image and return a vector of contours. Each contour consists of a series of points; Enclosing rectangle extraction: Once the contours are detected, the boundingRect function can be used to obtain the enclosing rectangle of each contour. This function returns a rectangle that defines the minimum enclosing rectangle of the contour.

[0022] Example 5 As a preferred embodiment of the present invention: in the extraction of the bounding rectangle, the center point, width, height and rotation angle of the minimum bounding rectangle of the contour are first calculated, and then this information is encapsulated in a Box2D object and returned. This minimum bounding rectangle can be used to describe the overall shape and direction of the contour; The bounding rectangle of the image after morphological processing is extracted, and the minimum bounding rectangle is calculated based on the geometric features of the contour. It first calculates the center point, width, height and rotation angle of the minimum bounding rectangle of the contour, and then encapsulates this information in a Box2D object and returns it. This minimum bounding rectangle can be used to describe the overall shape and direction of the contour. By setting a threshold on the area of ​​the bounding rectangle, the bounding rectangle whose area is greater than a certain threshold is retained, so as to extract the bounding rectangle of the tower.

[0023] Example 6 As a preferred embodiment of the present invention: in obtaining the inclination angle of the pole tower, based on the geometric analysis of the pole tower structure, the angle between the circumscribed rectangle side and the vertical direction is measured and calculated, thereby obtaining the specific inclination angle of the pole tower; The geometric structure analysis of the tower also involves the load calculation of the tower, the wind load calculation and the design of the tower foundation. The loads that the tower needs to bear include the weight of the hardware and insulator string, the weight of the conductor, the weight of the ground wire, etc. Specifically, the weight calculation of the hardware and insulator involves the sum of the specific models and quantities, the weight of the conductor is calculated according to the number and type of the conductor, and the weight of the ground wire also needs to be calculated according to the type and number of the ground wire. The wind load calculation of the tower involves dividing the tower into several sections. The wind load calculation of each section is based on factors such as the wind body shape coefficient and the wind pressure height variation coefficient. These calculations help determine the tower in a specific For stability under constant wind speed conditions, the design of the pole tower foundation needs to consider many factors, including geological characteristics, construction conditions, etc. The foundation types include pole foundation and tower foundation. The former mainly adopts prefabricated foundation, while the latter selects different foundation forms according to the tower type, topography and geology, such as original soil foundation, large excavation and backfill foundation, etc. The geometric structure analysis of the pole tower is a comprehensive process, involving load calculation, wind load analysis and foundation design. These analyses ensure the structural safety and stability of the pole tower, which can effectively support the transmission line and ensure the safety and efficiency of power transmission.

[0024] Example 7 As a preferred embodiment of the present invention: in judging the tilt state of the pole tower, according to the standard, a threshold is set, and the tilt angle of the detected pole tower is compared with other pole towers in the image. If the tilt angle difference exceeds the threshold, it is defined as the pole tower tilting, and if the tilt angle difference is lower than the threshold, the pole tower is not tilted; By performing image threshold segmentation on the strong scattering point information of the SAR image, and then using morphological methods to perform image segmentation, the contour information of the tower is extracted, and the inclination angle of the tower is obtained and calculated by extracting the circumscribed rectangle of the tower outline. Finally, through the relative geometric relationship between multiple towers in the same image, the inclined towers are identified and extracted, which saves manpower, material and financial resources to a great extent.

[0025] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations within the meaning and range of equivalents of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A tower tilt detection method based on high-resolution SAR images, characterized in that: include: SAR data set production: crop high-resolution SAR images to produce data sets, with each image containing multiple tower targets; Image preprocessing and binary segmentation: perform lee filtering on the image, use Otsu global threshold algorithm for binary extraction, classify the points with pixel grayscale values ​​greater than the threshold as the target, marked as 255 grayscale, and the rest as the background, marked as 0 grayscale, to complete the image segmentation; Morphological processing: The binary image is morphologically processed by using morphological opening operation, which first corrodes the image and then expands it to eliminate the influence of small noise points on the extraction of the tower, while retaining the shape characteristics of the tower as much as possible; Enclosing rectangle extraction: extract the enclosing rectangle of the image after morphological processing, and calculate the minimum enclosing rectangle based on the geometric features of the contour; Obtaining the inclination angle of the tower: construct the inclination angle of the tower through the angle between the longitudinal side of the circumscribed rectangle and the vertical direction; Determine the geometric relationship between poles and towers: mainly involving the height of the towers, the distance between the lines and the towers, the horizontal and vertical distances between the lines, and the arrangement of lightning conductors on the towers; Tower tilt state identification: By extracting the tilt angle of the tower, when the difference between the tilt angle of a tower and the tilt angle of other towers in the image is greater than a certain threshold, it is defined as a tilted tower.

2. The tower tilt detection method based on high-resolution SAR images according to claim 1 is characterized in that: The key technologies for high-resolution SAR image processing include motion compensation technology, imaging algorithm, image enhancement technology, and target recognition and extraction technology.

3. The tower tilt detection method based on high-resolution SAR images according to claim 1 is characterized in that: The lee filter is based on the principle of window function, and estimates the noise level by calculating the mean and variance of the pixel values ​​in the neighborhood around each pixel point, and performs filtering processing. The steps of the Otsu global threshold algorithm are: Preprocessing, grayscale histogram, threshold calculation, binarization.

4. The tower tilt detection method based on high-resolution SAR images according to claim 1 is characterized in that: In the morphological processing, the opening operation is an operation in the image morphological processing, which consists of two steps: first corrosion and then expansion. The purpose of the opening operation is to eliminate small noise points in the image while retaining the structural features of large objects in the image.

5. The tower tilt detection method based on high-resolution SAR images according to claim 1 is characterized in that: In the extraction of the bounding rectangle, the bounding rectangle in the SAR image is extracted by using the findContours method in the OpenCV library, which includes the following steps: edge detection, contour detection, and bounding rectangle extraction.

6. The tower tilt detection method based on high-resolution SAR images according to claim 1 is characterized in that: In the extraction of the bounding rectangle, the center point, width, height and rotation angle of the minimum bounding rectangle of the contour are first calculated, and then this information is encapsulated in a Box2D object and returned. This minimum bounding rectangle can be used to describe the overall shape and direction of the contour.

7. The tower tilt detection method based on high-resolution SAR images according to claim 1 is characterized in that: In obtaining the inclination angle of the pole tower, based on the geometric analysis of the pole tower structure, the angle between the circumscribed rectangle side and the vertical direction is measured and calculated, thereby obtaining the specific inclination angle of the pole tower.

8. The tower tilt detection method based on high-resolution SAR images according to claim 1 is characterized in that: In the determination of the tilt state of the pole tower, a threshold is set according to standard regulations, and the detected pole tower tilt angle is compared with other pole towers in the image. If the tilt angle difference exceeds the threshold, it is defined as the pole tower tilted. If the tilt angle difference is lower than the threshold, the pole tower is not tilted.

Images