球形全景图像的自动拼接

近年来,球形全景的虚拟现实方法开始发展起来.在已有的球形全景图像拼接方法中,基本上都是"半自动"的拼接方法,需要很多人工操作,效率低下、成本高昂.提出一种球面自寻匹配的拼接算法.该算法通过把多幅图像映射到一个合适的球面上,自动调整每幅图像的插入点,使得重合位置的差值图像灰度累积平均值最小,然后球面的图像反映射成平面形式,最终得到拼合图像.该算法可使程序实现完全自动化,使人工操作降至最低,从而降低全景图像的制作成本.

AUTOMATIC STITCHING OF SPHERICAL PANORAMA

Virtual reality method of spherical panorama has been developing in these years.Almost all of the known methods of spherical stitching are semi-automatic,which are inefficient and expensive due to much manual operations.This article introduces a sphere self-search and match algorithm.The algorithm automatically adjusts every image's inserting point by mapping multi-images onto a suitable sphere,to make accumulative average gray values of difference images on overlapped position be minimal,and then conversel...