改进的SURF彩色遥感图像配准算法

An Improved Surf algorithm for Color Remote Sensing Image Registration

摘要:

为了进一步提高彩色遥感图像的配准精度,针对遥感图像配准过程中色彩信息利用率低以及误匹配率高的问题,提出一种改进的SURF(Speeded Up Robust Feature)彩色遥感图像配准算法;该算法首先在对彩色遥感图像进行特征点检测基础上,对特征点描述算子进行改进,以使颜色空间变换后得到的特征点色彩信息添加到原描述算子中,并对特征点描述算子进行归一化处理,以增加算子的独特性和对旋转、尺度、光照的鲁棒性；其次,结合单向匹配和双向匹配的比值,提出了一种最优化阈值选择准则,如果欧氏距离比率小于最优化阈值,完成特征点匹配,得到正确匹配点对,再通过变换矩阵得到配准图像;实验结果表明,在保证实时性的条件下,该算法相比于原SURF算法,准确性和稳定性都有一定提高,具有一定的理论和应用价值。

Abstract:

To further improve the registration precision of color remote sensing image, an improved SURF registration algorithm is put forward aiming at solving the problem of low utilization rate of color information and high share of wrong matching pairs. Firstly, a modified descriptor is constructed adding in color information to increase the descriptor's distinctiveness and robustness. Secondly, combined with the ratio of one-way matching and two-way matching, an optimal threshold method is put forward to increase the accuracy. If the Euclidean distance ratio is less than the optimal threshold, the correct matching points could be acquired. And the registered image was obtained based on transformation matrix. Experimental results show, compared with the original SURF algorithm, that the improved algorithm has certain improvement in terms of accuracy and stability while guaranteeing the real time. It has certain values in both theory and practice.