一种仿生复眼系统的图像拼接算法研究

Image mosaic algorithm on bionic compound eye system

在研究昆虫复眼的基础上设计了一种仿生球面复眼系统,基于此系统提出一种图像拼接算法。首先采用SIFT(Scale Invariant Feature Transform)图像匹配算法进行各个子眼图像进行特征点提取;然后利用RANSAC(Random Sample Consensus)算法对提取出的特征点进行精确匹配;最后通过加权平滑算法实现多个子眼图像的无缝拼接,得到整个场景的全景图。仿真结果表明:算法可以有效地实现多通道图像的大视场拼接,增大多个方向的视场。该研究可进一步推进仿生复眼成像系统的应用。

Based on the compound eyes of insects, a bionic spherical compound eyes system was designed and an image mosaic algorithm was proposed. First, SIFT （Scale Invariant Feature Transform） was used to extract feature points from the images of sub-eye. Then, feature points were precisely matched by RANSAC （Random Sample Consensus） algorithm. Finally, seamless mosaics of the sub-eye images were completed with weighted smoothing algorithm. The simulation results showed that the algorithm could effectively realize multi-channel images mosaic with large field of view and enlarge the fields of multi-directions. The study could advance the application of the compound eye imaging system.