改进的SIFT图像匹配算法

Improved SIFT Image Matching Algorithm

在众多图像匹配算法中,SIFT算法是在总结了众多传统算法优势的基础上,将尺度空间理论融合到图像特征点提取过程中,这样SIFT算法就保持了图像的尺度和旋转不变性,另外在外部光照变化等因素的影响下也能对图像的特征点进行准确的匹配,但该算法在仿射变换方面还存在很大的不足。针对不足之处,采用SIFT算法的方法提取图像的特征点,然后通过使用ASIFT中的方法对提取到的特征点进行仿射变换以及为特征点分配方向,这样在增强了图像的抗仿射性的基础上也保持了图像的旋转不变性。实验结果表明,改进算法在保持了原SIFT算法各种优势的基础上,在增强图像的抗仿射性方面,可以取得良好的效果。

In many image matching algorithm, SIFT algorithm puts the scale space theory fusion into image feature point extraction process based on the summary of the advantages of many traditional algorithms, so that it keeps the scale and rotation invariance for image and in the external light illumination changes and other factors influence, can also match the image feature points accurately. But it stir exists great shortage in terms of affine transformation. Aiming at them,the SIFT method is used to extract image feature points and ASIFT method is applied to carry out affine transformation for extracted feature points and to distribute direction,enhancing image anti affine and maintaining the image rotation invariance. The experimental results show that the improved algorithm can achieve good results in enhancing the anti radiation of the image based on the advantages of the original SIFT algorithm.