中文摘要：

针对RANSAC算法迭代次数过多导致的图像拼接效率不高的问题,提出一种改进的RANSAC图像拼接算法;首先采用SIFT算法提取尺度不变特征点,利用双向互匹配策略对特征点进行匹配,在使用RANSAC算法计算单应性矩阵之前,利用相邻特征点之间的关系对初始特征点对进行筛选,最后使用加权平滑法完成图像的融合;实验结果表明该方法有效地减少了特征点对数,提高了RANSAC的运行时间,图像拼接效率有了很大的提高。

英文摘要：

In view of the problem that the excessive iterations lead to the bad effect of image mosaic, an improved RANSAC image mo- saic algorithm is proposed to solve the shortcoming. First, SIFT algorithm is used to extract scale invariant feature points. We use the way of interaction matching strategy to match feature points. Before calculate homography in use of the RANSAC algorithm, we select the initial feature points according to the relationship between the adjacent feature points. Finally, we use the weighted smoothing method to complete the image fusion. The experimental results show that this method is effective in reducing the number of feature points and improving the run- ning time of RANSAC, Image Stitching efficiency has been improved greatly.