영상 SLAM의 SURF 성능향상을 위한 MSAC/PROSAC 융합 알고리즘

초록

Camera has become a very important sensor in various areas, especially in visual SLAM(Visual Simultaneous Localization and Mapping), where one essential step is to extract matching points between two images and the next step is to remove outliers among the obtained matched points. RANSAC has become a standard method to remove outliers, and many RANSAC-releted algorithms have been developed, among which MSAC for accuracy and PROSAC for speed. In this paper SURF is used for matching and MSAC/PROSAC fusion algorithm is stuied for a fast and accurate algorithm. Performance analysis is done by using a camera in Apple phone 7, and two images are used to analyze the performance of RANSAC and MSAC/PROSAC.