An improved SIFT algorithm for feature points matching of dairy cow images

摘要

The conformation is the main parameter in dairy cow appraisal; hence, it is necessary for the measuring and evaluating of dairy conformation to use machine vision technology in 3D reconstruction. The feature points detecting and matching are the pivotal steps in 3D reconstruction. The conventional SIFT feature points matching algorithm has shortcomings of too large search area and low efficiency, An improved SIFT feature points matching algorithm is proposed to reduce the search range from both row and column directions based on binocular vision principle. 2.7 times of in speed by the improved algorithm has been shown in the experiments compared with that of the conventional algorithm with the same matching precision.