Abstract: Research on Scale-Invariant Features Transform(SIFT) feature match algorithm is done in order to make the algorithm more suitable to Unmanned Aerial Vehicles(UAV) images matching. Results of image match in different scales are analyzed and compared. A conclusion can be drawn that image match in reduced-scale costs shorter time but the match accuracy is still high. Image strategy combined with rough match and block match is proposed. Experiments prove this strategy is put out to get higher match efficiency.

Key words: Scale-Invariant Features Transform(SIFT) algorithm, scale, block-match

摘要： 为使尺度不变特征转换算法更适合无人机影像的匹配，对该算法及其匹配策略进行探讨。对不同尺度下的影像匹配情况进行分析比较，发现减少尺度的影像匹配可缩短计算时间，但结果精度并没有降低。提出粗匹配加分块匹配的影像匹配策略，通过实验证明该策略较整体匹配能提高正确匹配点对的获取效率。

关键词: 尺度不变特征转换算法, 尺度, 分块匹配

CLC Number: 

• TP18