Font Size: a A A

Study On Unmanned Aerial Vehicle Automatic Line-tracking

Posted on:2016-09-05Degree:MasterType:Thesis
Country:ChinaCandidate:T T HuangFull Text:PDF
GTID:2272330464467790Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
The electricity grid is developing rapidly all around the world. It brings safety maintenance a growing challenge. The inspection work of electricity grid has long been done by manual method. This method not only has low efficiency, but also has no ability to inspect the power lines in snowy weather or in the complicated terrain area. As drone technology developing, the trend of applying UAV to power lines tracking is irreversible. On the basis of line-tracking based on UAV visual, this paper studies the method of extracting power lines automatically from a image. The main contents include:a)The images sometime have unevenness degree of shade and noises. So this article studies the image preprocessing methods, firstly. Histogram equalization is used to enhance image, three filters are used to decrease noises. On this basis the paper studies five classical edge detection algorithms, and the edge detection method based on Morphology, Ratio edge detection algorithm and the wavelet edge detection method. Through comparison with the effects and run time, we can know that the edge detection method based on Morphology and the wavelet edge detection method more suited to follow-up study.b)After analyzing the developing process of the line extraction, the dissertation researches phase-grouping Radon transform and Hough transform emphatically, and then applies them to the line extraction. The power lines extracted by one of the methods alone can’t reach a satisfactory level according to experimental results. Considering Hough transform has good anti-noise and robust performance, this paper puts forward three algorithms to extract transmission lines. They are the directional wavelet transform based on Hough transform method, the Radon transform based on Hough transform method and the morphology based on Hough transform method. The results show that the three methods have all improved enormously on the accuracy and the integrity of power lines extraction, in addition, run time has been shorten in varying degrees.c)The power lines covered by other objects may not be a complete line during taken, or it is an approximate line with radians, not a rigid straight line because of gravity. So there are often circumstances that the extracted power lines are broken. The paper focuses on it, and analyzes the least square fitting method, which based on Hough transform and Kalman filter to connect power lines. The results show that the extracted power lines are more complete after being connected by the methods above.
Keywords/Search Tags:Morphology, Directional Wavelet, Phase-grouping, Radon Transform, Hough Transform
PDF Full Text Request
Related items