The Study On3D Reconstrustion Of Human Body With RGBD Sensor

3D human body reconstruction refers to the process of get human body3d informa-tion by using computer vision and graphics, it plays a crucial role in some applications, such as the3D movie,3D animation,3D games, virtual fitting and virtual reality, it is also one of the research hotspots of the computer vision and computer graphics. The traditional approach of the3D human body reconstruction requires higher consideration in scenes, generally, it requires complicated calibration. It exists some disadvantages such as low algorithm robustness, high computational complexity and time consum-ing, which lead to the poor user experience and not easy to popularize. In recent years, RGB-D camera such as kinect have attracted people’s attention, because it can obtain the object depth information quickly and effectively, many researchers use it for3D human body reconstruction, and made a lot of progress.Firstly, we narrated the researching background and significance of3D human body model reconstruction concisely in the first chapter. Then we introduce the prin-ciple of depth camera and registration algorithm often used in reconstruction of3D scanning in the second chapter.In the third chapter, we propose a easy3D self-portraits system that allows ordinary users to get fully textured3D models of themselves quickly and easily. Only a single Kinect sensor was used to acquire all the data needed, without the aid of equipment like rails and wheels. The entire process is completely automatic without any manual interaction. Compared to previous methods, our main contribution is simplifying the capturing process, in which only around eight single frames (with depth and color) at different views are required to reconstruct the entire model; In ad-dition, we propose an novel texturing method via texture mapping to color the model seamlessly. Finally,we summarize our work and prospect the future work in the lase chapter.