提出并实现了一种基于能量最小化的扩展景深方法,该方法使用对比度作为评价标准来拼接源图像,它的创新之处是利用了基于图分剖的能量最小化方法提供的空间一致性来解决拼接赝像问题,并用最大流算法来解决图分割问题。最后给出了实验结果,将实验结果与缺乏空间规范化的结果进行了比较。实验结果证明,本文方法能够在用户可以接受的时间范围内得到较为优越的扩展景深效果。
This paper proposes a kind of energy minimization based extended depth-of-field method, which uses a measure of contrast to compose source images. The innovation point is that it makes use of the spatial consistency afforded by the energy minimization via graph-cut to solve the problem of composite artifacts, and uses the max-flaw algorithm to solve the graph-cut problem. In the end of the paper, we present the result, and then compare it with the result of the algorithm without spatial consistency. The result shows that the method can produce more perfect effect within an acceptable execution time.