为提高影像匹配的稳健性，引入基于 SIFT 特征匹配的贝叶斯抽样一致性算法（Bayes sampl e consensus ， BAYSAC），提出基于随机概率U （0，1）、基于像点到像片中心距离比值和基于影像重叠度的3种正确点先验概率估计方法，并根据相似性原理简化了贝叶斯公式，用于更新正确点概率。以SIFT算法为基础，结合贝叶斯抽样一致性算法，对不同的正确点概率估计方法进行了试验。试验结果表明，改进后的算法减少了迭代次数，从而减少了计算时间。同时，它能剔除更多的误匹配，并保留更多的正确匹配，从而提高匹配正确率。%In order to improve the robustness of image matching , an image matching method based on SIFT and Bayes sampling consensus is introduced .Three methods are proposed ,prior inlier probability estimations based on random probability ,ratio of distances between image point and image principal point and image overlap .The inlier probabilities with simplified Bayes rules are updated after each test .Combining SIFT algorithm with BAYSAC ,the experiment results of different prior inlier estimations are shown .The results show that with BAYSAC algorithm ,the number of iterations and the computational cost are reduced .BAYSAC can remove more outliers and save more inliers than RANSAC , which improves the correct rate of matching and thus proves the validity of the proposed al gori thm .
展开▼
