一种遥感图像高精度自动拼接算法

An Automatic and High Precise Mosaic Algorithm for RS Image

图像的自动拼接在地理信息系统和全景视频系统中应用广泛,但是传统的自动拼接算法在拼接含有大片相似区域的图片时精度不高,效果不理想.本文在传统算法的基础上,提出一种高精度的自动拼接算法,利用基于特征的算法提取特征点,利用基于像素的算法寻找匹配块,然后根据RANSAC原理精确计算图片的匹配关系.实验证实,该算法在图片包含大量相似区域及重叠部分相对较小的情况下,有比传统算法更高的拼接精度,而计算量并未增加.

Automatic image mosaics are widely used in GIS and Panoramic Video System, but traditional algorithms can＇t cope well with the images which contain a lot of similar areas. In order to solve this problem, we propose a new algorithm based on the conventional algorithms. First, we extract the feature points. Second, looking for the matching block, and finally we obtain the point transformation parameters between two images using robust RANSAC algorithm. The experiments show that our algorithm can perform more effectively than traditional algorithms when the images have larger similar areas and less overlap.