A Stereo Matching Method Combining Feature and Area Information for Power Line Inspection

Since the stereo vision has the advantages of providing autonomous navigation information for unmanned aerial vehicle (UAV) without environmental constraints. In this paper, the UAV is combined with stereo vision to perform the power line inspection. With the stereo vision system the line segments are reconstructed to obtain the orientation information, which provides reference for the attitude adjustment of the UAV. During the line construction, in order to make the line inspection more natural and reasonable, an efficient stereo line matching method which combines feature matching and block matching is proposed with artificial screening. Firstly, the image returned by the on-board binocular camera is used to click on the specific line to eliminate interference in the clutter background. The region of interest is filtered in the Hough space of the left image, and the Iterative Closest Point algorithm is used to filter the matching area in the right image. Secondly, The SAD matching cost function is used to merely recover the disparity map of the region of interest for the stereo frames after line screening. The reconstruction of a particular line is performed by least squares. To the performance of the proposed approach, the cyclic experiments on the ground are carried out. Based on the results, the proposed method improves the angle measurement accuracy of line segments. The orientation information can be utilized to provide reference for the attitude adjustment of the UAV.