摘要
针对无人机遥感图像畸变较大,而传统快速鲁棒(Speeded-Up Robust Features,SURF)算法不能提供足量兴趣点的问题,提出了一种基于Harris角点和SURF算法的无人机遥感图像配准方法。首先构建多尺度空间,并在多尺度空间下检测Harris角点作为兴趣点;然后计算各兴趣点的64维SURF描述子;最后运用K-d树匹配搜索策略得到两幅图像的匹配点对。将该方法与传统SURF配准方法进行实验对比,实验表明改进算法在保证实时性的情况下可以获得更多的匹配点对,并具有更高的配准精度。
According to the problem of large distortion of UAV Remote Sensing Image, while the traditional fast Robust （Speeded Up Robust-the Features,SURF） algorithm cannot provide enough feature point.This paper proposes a registration .algorithm based on the Harris corner with SURF description.First,the scale spaces were built,and the scale-invariant Harris corners are selected as feature points.Then,calculate the 64-dimensional SURF descriptor of every corner.Finally,feature points are matched by the k-d tree search strategy.The results of this way have comparison with the results of traditional SURF algorithm.
出处
《工业控制计算机》
2014年第7期135-136,139,共3页
Industrial Control Computer
基金
国家自然科学基金(61073196)
陕西省教育厅专项基金资助项目(08JK319)
陕西省科学技术研究发展计划项目(2011K17-04-01)
