基于分块和仿射不变性的SIFT图像匹配算法

针对传统SIFT算法在匹配时出现实时性差、匹配量低以及RANSANC算法在剔除SIFT误匹配对时误匹配率高的问题，提出一种基于距离相对性的分块匹配算法和基于仿射不变性的误匹配对剔除算法。首先利用传统SIFT算法提取图像中的特征点；然后采用基于距离相对性的分块匹配算法进行特征匹配得到初始匹配对；由于初始匹配对中存在误匹配，接下来运用基于仿射不变性的误匹配对剔除算法来剔除误匹配对；最后，在不同图像变换下进行仿真实验。实验结果表明，算法在保持SIFT算法鲁棒性的基础上，能够得到更多匹配对，正确匹配率提高了10%左右，并且实时性也得到很大改善。

SIFT Algorithm of Image Matching Based on Block and Affine Invariance

This paper proposed a block matching algorithm based on distance relativity and mismatch pairs elimination algo-rithm based on affine invariance, in order to solve the problems of bad real time capability and low matching of the traditional SIFT algorithms and the high mismatch rate when RANSANC algorithm eliminate the mismatch pairs. Firstly, using the traditional SIFT algorithm extract the feature points in the image . Then, the initial matching pairs are gotten by the block matching algorithm based on distance relativity. Because of the mismatch pairs in the initial matching pair, then mismatch pairs elimination algorithm based on affine invariance is used to remove the mismatch pairs. Finally, simulation experiments are carried out under different image transformations. The experimental results show the algorithm can obtain more matching pairs on the basis of maintaining the robustness of SIFT algorithm, the correct match rate increased by about 10%, and the real-time performance has also been greatly improved.