An Optimized Model of Polyhedral Visual Hull Based on Feature Points Matching

The 3D reconstruction based on the video image in real-time is an important problem, such as computer vision, computer graphics. The accuracy of the model is even the center issue of these fields. After considering the polyhedral visual hull (PVH) reconstruction method, this paper introduces a model optimized method based on the image feature points matching. In this paper, we present an algorithm that extracts the feature points from the public visible field of neighbor images, and finds the matched feature points by utilizing the epipolar relationships and the photo-consistency, and optimizes the original model by utilizing the matched feature points achieved and the original space location. The experimental results show that this method optimizes the original model efficiently and make the optimized model more accurate.