Strategies of improving matching accuracy about SURF
Quan Wei, Qiao Liu, Han Cheng, Xue Yaohong, Hua Li · 2017
Speeded-Up Robust Features(SURF) algorithm is a very popular matching method which are often used in fields such as feature extraction, fast image matching, depth map calculation and so on. About this algorithm, researches mainly focuses on improving the matching accuracy without increasing the matching time. In this paper, we mainly study on SURF algorithm, and based on which three kinds of improving strategies are discussed. They are Lowe optimization algorithm, the descriptor dimension extension algorithm and the direct optimization method. Experimental results show that Lowe optimization method has the highest accuracy under different situations, and the descriptor dimension extension algorithm will be a good choice too, if the mismatching points can be removed effectively.