一种改进的基于形变方法的图像拼接算法

Improved image mosaic algorithm based on morphological method

为实现存在形变和曝光差异的图像之间的无缝拼接,提出一种改进的基于形变方法的图像拼接算法。该算法对已经配准的待拼接图像的重叠区域进行图像融合,先利用直方图匹配法使两幅待拼接图像分别生成符合对方曝光程度的图像,然后利用基于水平集的序列图像生成算法得到在两种曝光程度下的融合图像,再结合多分辨率图像拼接算法和动态规划最佳缝合线查找法对得到的融合图像进行拼接,最终获得拼接图像。通过对合成图像和标准数字图像的模拟拼接实验表明,该算法能同时实现图像形变和曝光差异的平滑过渡。

In order to achieve seamless mosaic image with both shape and exposal differences,an improved image mosaic algorithm is proposed based on the morphological method.In this algorithm,the image fusion is applied on the overlapping region of registered images.Firstly,the histogram matching method is used to generate two other images with the histograms in accordance with each other.A level-set approach to image blending is then applied to obtain blended images under two different exposure amounts.The final mosaic image is attained by stitching these two blended images using a combined algorithm of multi-resolution mosaic approach and the optimal seam line searching method based on the dynamic programming.With simulating mosaic experiments on both synthetic images and standard digital images,this algorithm is proved to be capable of achieving smooth transition between images with both shape and exposal differences.