Visual Hull Construction Based On Image Sequences

In recent years, marker-less motion capture has been a hot topic in the fields of computer graphics and computer vision, and it has been widely used in a variety of fields such as in computer animation, games, stunt. Marker-less motion capture system achieved by Visual hull, which is constructed from image sequences, can apply to the simulation of human motion, biomechanics analysis and virtual reality in which have greater demands in the convenience and security.The silhouette is very important information. It can construct the visual hull of 3D model from a set of image sequences by computing shape-from-silhouette. While because of the difficulties of the 2D-3D projection, occlusions and self-occlusions, the current visual hull construction based on image sequences are still a lot of technical difficulties. This dissertation proposed a new visual hull generation algorithms, meeting the need of the motion capture data and improving the accuracy of data. The main study works and achievements include following aspects:1. It introduces the research background and purposes of the image-based modeling, summing up the typical image-based modeling algorithm and detailing description of the relevant theoretical knowledge of the image-based modeling.2. In this paper anew visual hull algorithm was represented. Compared with traditional methods, our approach employs the maximum overlap line segments intersection to calculate line segment intersection stably. In addition, the extraction of the border surface is more accurate. The experimental results demonstrate the algorithm efficiency, robustness, results close to the original model.3. A visual hull prototype system is developed; the important technologies are verified and applied in project. Reconstruction results have realistic. User and systems can interact to carry out a variety of space operations to achieve the 3D reconstruction of the object.