基于改进的SIFT图像匹配方法研究

Study on the Improved SIFT Image Matching Method

对SIFT(Scale Invariant Feature Transform)算法进行了研究。针对高斯差分金字塔阶数与特征点数关系问题,对算法进行了改进,通过控制金字塔的阶数来控制特征点数。实验数据分析表明:在图像存在不同程度的旋转变形和噪声影响时,改进后的算法鲁棒性更好,实时性更强。

Study on the SIFT （Scale Invariant Feature T ransform ） algorithm ,to solve the problem of difference of Gaussi-an pyramid order and feature points relations ,improves the algorithm ,to control the feature points by controlling the or-der of pyramid .After analysis of the experimental data ,indicates when the image exists different degrees of rotation distor-tion and noise impact ,improved algorithm has better robustness and better real-time .