A Mosaic Algorithm of Panoramic Image Based on Particle Swarm Optimization and Mutual Information
Lingda Wu · Jisuanji fangzhen · 2009
Constructing a virtual space by using the technique of Image-Based Rending(IBR) has now become a hot topic in the research field of virtual reality,and panorama image mosaic is one of the important research contents of IBR,in which a 3D scene is constructed by using image-composing instead of 3D geometry modeling.In this paper,a mosaic algorithm of panoramic image based on particle swarm optimization(PSO) and mutual information(MI) is proposed.Firstly,the algorithm employs the PSO algorithm to search template area including enough features,and then it makes use of mutual information in panoramic image mosaic.The algorithm is able to seamlessly mosaic a group of images automatically,no manual selection of feature points is necessary.The experiments show that it can be applied in panorama image mosaic successfully.It not only can significantly improve the speed of rending,but also can efficiently reduce negative effect introduced by circumvolution,which is demonstrated to be a rapid and practical algorithm.