摘要
针对图像拼接过程产生的接缝问题,采用最佳接缝选择算法缝合图像接缝。然而合适的能量函数是最佳接缝选择算法的关键。提出一种基于Canny边缘检测算子改进的能量函数,其目的是最小化图像重叠区域的梯度差异。通过实验将提出的能量函数与亮度误差、梯度差、梯度和、梯度差和梯度和的加权和以及Canny边缘检测算子等五种常用能量函数的最佳接缝选择结果进行对比。结果表明,提出的能量函数可以减少拼接图像的结构接缝,而且对于光照变化具有更好的分割稳定性,使得拼接之后的图像看起来更加自然美观。
Concerned the seams problem arising from image stitching, an optimal seam selection algorithm is used. And an appropriateenergy function is the key to the algorithm. An improved energy function based on Canny detector is proposed in this paper. Its goal is tominimize the differences of the gradients between overlapped areas of the images. Moreover, the proposed energy function is comparedwith the other five common ones by experiments, which include the brightness errors, the gradient errors, the gradient sums, the weightedsum of the gradient errors and sums, and Canny edge detector. The results show that the proposed algorithm can effectively reduce thestructure seams, and has robust segmentation against changing illumination and it makes the mosaic images more natural.
出处
《武夷学院学报》
2016年第6期65-69,共5页
Journal of Wuyi University
基金
莆田学院育苗基金资助项目(2014057)
