科研之友
微信
新浪微博
Facebook
分享链接
摘要
This paper proposes a multicamera setup-based stitching scheme for generating wide field-of-view (FOV) video without ghosts. Wide FOV videos can be constructed using an extensible wide FOV model for multiple cameras and a stitching process. With the extensible wide FOV model for multiple cameras, video streams are captured synchronously from multiple cameras without the need for a specialty designed lens; viewing angle of wide FOV video can be up to 360 degrees. Using the stitching process, a wide FOV video without ghosts can be constructed from the video streams. We propose matching the appropriate corresponding pairs based on the corner,; in the overlapping region of two adjacent images. This greatly reduces the computational. cost and improves the accuracy of the corresponding pairs. We also propose a two-phase image blending scheme to obtain a seamless image. The scheme, which uses half of the image, can alter a wider range of color intensity than is possible if only the overlapped part of the image is used. Therefore, it can smooth the color transitions from one image to another. The proposed two-phase image blending scheme balances the significant intensity difference in adjacent images, and smoothes the color intensity of overlapping regions in adjacent images. Experimental results show that the proposed method performs well for generating wide FOV videos.
- 出版日期2009-11
