Feature Extraction and Matching of Slam Image Based on Improved SIFT Algorithm

Abstract

In order to improve the robustness and accuracy of slam system, the Improved SIFT algorithm is used to extract the image features. Firstly, the characteristics of the image in slam are analyzed and the image preprocessing is carried out to reduce the gray mutation. Secondly, in order to meet the real-time requirements, the feature descriptors of sift are simplified to improve the speed. Using the continuity of slam image, the method of pixel neighborhood matching reduces the time of feature matching and reduces the error matching rate of repeated texture. GPU is used to implement the Improved SIFT feature algorithm. Finally, the simulation results show that the trajectory accuracy is improved by more than 35% and the image processing time is about 12ms. At the same time, the system accuracy is improved.