Research on ORB feature point extraction algorithm based on adaptive threshold
In weak texture scenes,in the process of extracting feature points with the ORB algorithm,fixed threshold detec-tion of FAST corner points may lead to poor extraction results and affect the matching accuracy.This paper proposes an adaptive threshold ORB feature point extraction algorithm,which uses image grayscale The difference value and pixel distribution probabil-ity are used to calculate the image contrast,and the corner detection threshold is dynamically calculated based on the contrast.Then the feature points were extracted based on the dynamic threshold algorithm,using two matching algorithms:brute force match-ing algorithm and fast nearest neighbor matching(FLANN).The feature point matching accuracy and accuracy of the ORB algo-rithm,SIFT algorithm and the algorithm in this paper were compared on the EuRoc data set.time consuming.The results show that the matching accuracy is 26.6%higher than the ORB algorithm and 32.7%higher than the SIFT algorithm.
ORB algorithmadaptive thresholdfeature pointsvisual SLAM
NETL
NSTL
万方数据
维普
