基于SURF的序列图像快速拼接方法

摘要

针对车载摄像平台的序列图像拼接问题,提出一种基于SURF的序列图像快速拼接方法.首先,对经预处理后的序列图像提取SURF特征点,采用最近邻法匹配特征点;其次,利用RANSAC算法估计相邻帧的变换关系;再采用改进的全局配准策略,完成图像配准;最后,采用一种最大值融合法,去除了拼接缝和光照明暗变化的影响,实现序列图像的无缝拼接.实验结果表明该方法适用于车载摄像平台序列图像的快速、无缝拼接,具有较强的鲁棒性,有一定的实用价值.%A fast method based on SURF is proposed for image sequences mosaic of vehicle camera platform. First, SURF features are extracted from preprocessing images and matched by using the nearest neighbor method. Second, RANSAC is used to solve for the homography between adjacent frames. Then, a global model is found with improved image registration method. Finally, seamless image sequences mosaic is obtained with image fusion of maximum, which is against seams and illumination variations. The experimental results show that the method is fast for seamless image sequences mosaic of vehicle camera platform, is robust and has certain practical value.