Abstract:

In order to reduce the time consumption of UAV video image dense matching, and improve processing efficiency, this paper presents a fast feature description and feature matching method. The method uses Harris operator to detect corners, adopts feature descriptor S-DASIY (DASIY) to characterize detected corners and generate the 25-dimensional feature descriptor for the corners. The feature points of the images are matched in accordance with the appropriate matching criteria, to get the match points between images. Experimental results show that the proposed fast feature description and feature matching method can significantly reduce the time of generating and matching feature points.

Key words: UAV, dense matching, feature descriptor, video images

摘要：

为了降低无人机视频影像密集匹配的消耗时间, 提高处理效率, 提出快速的特征描述和特征匹配方法。该方法首先利用Harris算子检测角点, 然后通过特征描述符simplified-DASIY(S-DASIY)对检测角点进行特征描述, 生成该点的25维特征描述符。根据相应的匹配准则对影像对特征点进行匹配, 得到影像对的匹配点。实验表明, 该方法能够显著减少特征点的生成和匹配时间。

关键词: 无人机, 密集匹配, 特征描述符, 视频影像