| abstract |
The invention provides a feature matching method for binocular image stitching of a mobile inspection robot. In the feature point detection stage, a scale pyramid is constructed and feature points are extracted using the extremely fast FAST algorithm to enhance the robustness of scale invariance; The improved CS-LBP description method describes the feature points, enhances the robustness of rotation invariance, and reduces the dimension of the feature vector to improve the matching efficiency; finally, the DDRN algorithm is used to measure the similarity of the feature vector to complete the matching, and the improved RANSAC algorithm Eliminate false matches. Compared with the traditional algorithm, this algorithm has greatly improved the real-time performance and accurately realized the feature extraction and matching of the image, and the improved description method has strong anti-interference ability for the rotated image, and it can be used in complex fields such as affine, scaling, and lighting. Changing scenes still has high adaptability. |