基于全局最优变换矩阵的图像拼接方法

Image Stitching Based on Global Optimal Homography Matrix

针对目前图像拼接在重叠区域出现重影、误对齐,非重叠区域的透视失真问题,提出了最优单应性矩阵的视差图像的图像拼接算法。首先根据待拼接图像的特征点分布和匹配点的对齐误差获得最佳全局单应性矩阵;再利用RANSAC算法计算全局最优相似变换矩阵,并结合最优单应性矩阵获得输入图像的形变图像;最后利用最佳拼接缝搜索算法获取最优拼接缝,加权融合图像;实验表明:基于最优矩阵的拼接算法有效避免穿过重叠区域的重要目标,减缓非重叠区域的透视失真,拼接效果更加自然真实。

Aiming at the problems of ghosting,misalignment in overlapping areas and perspective distortion in non-overlapping areas,an image stitching algorithm for parallax images with optimal homography matrix was proposed.First,the best global homography matrix was obtained based on the feature point distribution of the images to be stitched and the alignment error of the matching points;Then the RANSAC algorithm was used to calculate the global optimal similarity transformation matrix,and the optimal homography matrix was used to obtain the deformation image of the input image;The optimal stitching seam search algorithm was used to obtain the optimal stitching seam and weighted fusion image.Experiments show that the stitching algorithm based on the optimal matrix effectively avoids the important goal of passing through overlapping areas,slows down the perspective distortion of non-overlapping areas,and the stitching effect is more natural and real.