针对线性加权平滑法找到的拼接缝两边的灰度差异较大,存在缝较明显的问题,提出最佳拼接缝的线性加权平滑法。该算法首先从图像重叠部分的中间区域开始选择每行梯度和最小的像素作为最佳拼接点;同时,对于图像受噪声的影响,以及存在多个像素点最小"梯度和"相等的问题,利用散度量进一步检查最佳拼接点,然后连接每行的最佳拼接点作为最佳拼接缝,最后使用线性加权平滑法消除拼接缝。试验结果表明,此方法使图像重叠区域能平滑自然过渡,实现无缝拼接。
As the linear weighted smoothing method causes obvious seam-line due to large gray difference between the two images, a linear weighted smoothing algorithm for the optimal stitching-line removal is proposed. The algorithm first selects each line gradient and the smallest pixels from the middle of the overlapping area of the image as optimal stitching point, and as for the image affected by noise and the problems of multiple pixels minimum "gradient sum" equal, uses divergence metric to check the optimal stitching point, then connects optimal stitching point of each line as a best seam, at last applies linear weighted smoothing algorithm to remove stitching line. The experimental result shows that the proposed method achieves image overlap region smooth transition and seamless stitching.