【摘要】 文章针对图像自相似或具有对称性时SIFT匹配稳定性不高的问题,研究基于Harris尺度不变特征的图像匹配方法。为了获取更稳定的特征点,将SIFT特征描述方法引入到Harris尺度不变特征描述中,改进了基于Harris特征的匹配算法;结合简单高效的基于欧氏距离的双向匹配算法,去除了大部分的错误匹配,明显提高了匹配的稳定性。实验结果表明,改进的算法不仅对图像具有平移、旋转以及尺度不变性,且对于具有自相似或对称性的图像之间的匹配稳定性更高。更多还原

【Abstract】 This paper focuses on image matching algorithm based on Harris scale invariant feature in view of the fact that scale invariant feature transform(SIFT) matching efficiency is low when the images are self-similar or symmetric.In order to obtain stable feature point,the SIFT method is applied to describing Harris scale invariant feature,which improves the matching algorithm based on Harris corner.By using simple and efficient bilateral matching algorithm based on Euclidean distance,most mismatches are removed and the robustness of matching is improved.The experimental results demonstrate that this improved algorithm achieves movement,rotation and scale invariant in matching images,and has a higher registration precision on images which are self-similar or symmetric.更多还原