Research on Feature Detection and Matching Based on Improved ORB Algorithm
This paper proposes an improvement to the ORB feature extraction and matching algorithm to address the shortcom-ings of the traditional version,which is susceptible to errors due to its reliance on a fixed queue value,sensitivity to lighting condi-tions,limited ability to extract texture information,limited ability to extract texture information,and a tendency for feature point en-richment.Firstly,the region is divided with the point to be detected as the centre.Secondly,an adaptive threshold is used to judge the detection of the point,and an adaptive range non-maximum suppression algorithm is applied to address the feature point enrich-ment issue,thereby enhancing uniformity.Thirdly,the texture information extraction ability is improved using various anisotropic diffusion filtering.Finally,the descriptor pairs are improved by combining bilateral filtering,and the mismatch is eliminated by the PROSAC algorithm.The experimental results demonstrate that the enhanced algorithm exhibits a robust capacity to adapt to varying lighting conditions,enhances the capability to extract texture features,ensures a more uniform distribution of feature points,im-proves the enrichment effect,and enhances the accuracy of matching under light conditions.It displays superior adaptive capabili-ties.
国家科技期刊平台
NSTL
万方数据
