Research On The Line Feature Matching In Stereo Image Based On The Main Points Of The Edge

Stereo image matching is a core issue in the field of digital photogrammetry, the essence of which is to identify the same points between the two (or more)images,it widely used in the three-dimensional reconstruction,computer vision,target recognition and so on.Characteristic lines as one of the most notable and important features in the sense of sight, it generally exist in the real feature and target scenes.Characteristic line was chosen as the preferred matching primitives in the image matching technology because of it contains rich and stable features.Therefore,seeking a strong applicability,high precision matching algorithm is an urgent task.A characteristic line matching method under the action of the main points of edge is proposed,which transformed the line matching to the main points matching.This method strive to achieve the line matching with the "points substitute lines" and "straight lines substitute curves".The main contents of this dissertation are as follows:(1) Image pre-processing.For the noises and low contrast areas in the stereo image,this dissertation adopts the adaptive smoothing filter to reduce the image noise and the Wallis filter to enhance the image features.This method highlighting the existing features and improve the accuracy of feature extraction in order to realize the efficient and reliable foundation for image matching.(2) Extract the line feature and the main points of the edge.On the basis of analyzed the existing algorithm of the line extraction,this dissertation adopts a line extraction algorithm based on chain code standards.This algorithm achieved the object edge extraction by the edge detection,the chain code tracking and the chain code division based on eight direction chain code standards.The proposed algorithm can not only extract the straight edge of the objects, but also can use straight segment to approximate the boundary of the irregular objects efficiently and accurately.(3) Line matching.A characteristic line matching method under the action of the main points of edge is proposed.Firstly,use the SIFT operator and RANSAC algorithm to get the same points between the stereo image in order to obtain the homographic matrix.Secondly,adopt the homographic matrix constraint,the epipolar ray constraint and the gray-scale similarity constraint to search the same points of the main points of edge.Finally,connect the main points of edge which was matched successfully,eliminate the false maches by the visual interpretation to achieve line matching.