改进SIFT算法在图像数据相似性匹配中的应用研究

Application Research on Comparability Matching for Image-Data Based on Enhanced SIFT Algorithm

针对基本SIFT算法在效率和准确性上的不足,通过减少提取特征点主方向以及降低在生成描述子时的重叠运算复杂度,从行列两方面缩小特征点匹配的搜索范围,提高了算法的执行效率,增加了相似性匹配的匹配准确率,使其在月球图像数据库匹配的应用中取得了较好的效果。

For the deficiencies on efficiency and accuracy of the SIFT(Scale Invariant Feature Transform)algorithm,improvement was made by two ways,in which one is to reduce the overlap computation on major direction and generation of description,and the other is to shrink the search coverage of feature point match.The experiment on the match of moon image database gives a preferable result,in which the real-time performance and matching accuracy are improved.