Feature Matching Algorithm Based on SURF and Lowes Algorithm

Abstract

Feature extraction and matching of images is a key step in 3D reconstruction, and its accuracy directly affects the accuracy of 3D reconstruction. In this paper, aiming at the mismatch caused by the high similarity between screws, proposes a feature matching algorithm based on median filtering, Lowes algorithm and scale-invariant feature transform (SURF), called M-L-SURF algorithm. First, the median filtering is performed on the screw image to remove noise, then the SURF algorithm is used for feature extraction and matching, and finally, the Lowe's algorithm is used to filter the matching results. The results of experiments show that the M-L-SURF algorithm can achieve a 97.4% correct rate of screw image matching. The matching results obtained in this paper can be better applied to the subsequent work of 3D reconstruction.