A Fast Identification Method of Transmission Lines Based on Partial Derivative Distribution and Boundary Strategy

Abstract

The invention proposes an accurate and real-time transmission line identification method, including two innovations: a new integral formula of Radon transformation and a search method of Radon transformation based on boundary strategy. The new integral formula of Radon transformation proposed by the present invention can replace the traditional target enhancement process, and the value of Radon integral transformation in the background part is suppressed through the symmetry of partial differential function values, while the transmission line part is reserved. While proposing a new integral formula of Radon transform, the present invention starts Radon transform from the boundary of the image according to the requirements of the application, and limits the search area by selecting the boundary, thereby achieving the purpose of reducing calculation time. The symmetric distribution law of the natural image partial differential function used in the present invention is first discovered in this project, and there is no related research at home and abroad. Through theoretical research and experimental verification, the invention can effectively identify power lines in various complex aerial images in real time.

Description

technical field [0001] The invention relates to digital image processing technology and pattern recognition, in particular to an automatic identification method for transmission lines applied to the inspection of high-voltage transmission lines by drones. Background technique [0002] It is of great practical significance to carry out regular inspections on high-voltage transmission lines to ensure the safe operation of the power grid system. Early monitoring of transmission lines mainly relied on manual line inspection, which was greatly affected by terrain environment and weather conditions, and had disadvantages such as low efficiency, heavy workload, and long cycle. Using unmanned helicopters to inspect power transmission lines is a new technical means, which has the advantages of high efficiency, low cost and wide application range. The use of vision-based automatic identification technology for transmission lines can assist UAVs in autonomous line inspection, greatly ...

AI Technical Summary

Problems solved by technology

The eigenvalues ​​and eigenmatrix of the Hessian matrix have different characteristics in different image structures. Literature [5] uses the analysis of the Hessian matrix to find linear targets in the image. This method can better strengthen the characteristics of high-voltage lines. However, its calculation time is long and its real-time performance is poor.

[0008] (4) Some researchers use the method of image segmentation to enhance the transmission line. Literature [6] proposed a 2-dimensional maximum inter-class variance method based on simulated annealing particle swarm optimization algorithm for image segmentation. The problem of this method is Poor real-time performance

[0010] (5) Hough and Radon transforms are common methods for detecting straight lines. Most of the current literature uses these two methods to identify transmission lines [3-5]. Among them, literature [5] adds angles to the radon transform. Constraints, the literature in [3] uses parallel line constraints to identify transmission lines after Hough transform, the recognition rate of this method is low, usually only part of high-voltage transmission lines can be identified

[0012] Judging from the methods and methods listed above, the current automatic identification methods for transmission lines are all carried out on the basis of transmission line enhancement, and the final identification effect is greatly affected by the enhancement effect of linear targets.

In the case of complex background, if the power line fails to be enhanced from the background image, it will lead to the failure of the subsequent recognition process

In addition, the traditional Hough and Radon transform, as a kind of ergodic integral operation, has a very large amount of calculation, which affects the real-time performance of the algorithm.

In practical applications, the current methods have many inherent disadvantages that need to be improved

Images