平行多视点算法求取深度信息

Real-Time Implementation of Calculating Depth Information by Image Sequences Acquired from the Parallel Multi-Viewpoints

引入了平行多视点的概念，从平行多视点获取的多幅图像中可以求得深度信息。通过利用图像的ＦＦＴ变换，将时域中很难解决的问题变换到频域中来解决，避免了复杂的匹配运算，极大地简化了求取图像点、面深度信息的问题，并且提高了深度求取的实时性、全面性和准确性。介绍了算法原理及特点，并给出了实验结果及分析。

A new algorithm of calculating depth information by image sequences acquired from theprallell multi-viewpoints is discussed. By means of orthogonal transformation of image(in this paper,authoruses fast Fourer transformation),this paper puts the matching problem which is difficult to solve in timedomain into transformation field in frequency domain. This method greatly simplifies the problem ofacquiring point and surface depth information,and its mathematics deduction as well.Then the authors statethoroughly the principle of the algorithm and characteristic of acquiring depth information by parallel multi-viewpoints. At the end of this paper,experimental results including simulation experiment results andexperirmental results of practical photographing image are given.