基于鲁棒迭代优化的图像拼接算法

J4 ›› 2014, Vol. 36 ›› Issue (04): 741-745.

基于鲁棒迭代优化的图像拼接算法

杜培明 1，邢硕 1,2

（1.安徽工业大学电气信息学院，安徽马鞍山 243032;2.铜陵学院电气工程学院，安徽铜陵 244061）

收稿日期:2012-10-24 修回日期:2013-01-21 出版日期:2014-04-25 发布日期:2014-04-25

Image mosaic based on robust iterative optimization

DU Peiming1，XING Shuo1,2

(1.School of Electrical Engineering & Information,Anhui University of Technology,Ma’anshan 243032;2.School of Electrical Engineering,Tongling University,Tongling 244061,China)

Received:2012-10-24 Revised:2013-01-21 Online:2014-04-25 Published:2014-04-25

摘要/Abstract

摘要：

针对保持直线边缘不畸变和改善拼接区域精度的需要，提出了一种基于鲁棒迭代优化的图像拼接算法。采用SIFT特征匹配算法提取不变特征，依据RANSAC鲁棒算法估计单应性矩阵获取精确内点匹配点对，参照内点匹配点对作为参数，利用最小二乘法重新估计单应性矩阵，最终通过鲁棒迭代优化重新计算更多内点直至内点收敛。实验结果表明，所提出的算法使得直线边缘维持直线化，同时拼接精度得到提高。

关键词: SIFT特征匹配, RANSAC鲁棒估计, 最小二乘估计, 迭代优化

Abstract:

Aiming at the requirements of keeping edges in a straight line and improving the mosaic precision, an image mosaic algorithm based on robust iterative optimization is proposed.Firstly,SIFT feature matching algorithm is employed to extract features with invariance. Secondly, inlier matching points are obtained accurately through the homography matrix that is estimated according to RANSAC robust algorithm. Thirdly, the homography matrix is reestimated utilizing least square method by treating inliers as parameters. Finally, more inliers are recomputed until converge based on robust iterative optimization. Experimental results show that, with the proposed method, edges are maintained as a straight line and the precision of image mosaic is greatly improved.

Key words: SIFT feature matching;RANSAC robust estimation;leastsquares estimation;iterative optimization

杜培明 1，邢硕 1,2. 基于鲁棒迭代优化的图像拼接算法[J]. J4, 2014, 36(04): 741-745.

DU Peiming1，XING Shuo1,2. Image mosaic based on robust iterative optimization [J]. J4, 2014, 36(04): 741-745.