Rapid3D Human Body Modeling And Skinning Animation Based One Kinect

Rapid3D human body modeling technology is one of the important research issues innowadays computer vision (CV), virtual reality technology and computer graphics, which isalso the developing tendency of the model’s fast reconstruction.3D object’s reconstructioncan be cheaper as well as accessible like photos and video, for the reason above,3D modelcan be applied into more area, such as E-commerce platform, Online shopping sites, andcomputer games and so on. Thus, how to model human’s body fast and cheaply is theimportant goal to be achieved in CV, virtue reality technology and computer graphics.However, to the temporary approach for the high quality human body’s model, the equipmentis not only expensive but also inconvenient to scan body.To make thing worse, equipment is always accompanied with complicated operation andthe long time production. All of the defects above limit the application of model3D humanbody, especially to the virtual reality system that requires high instantaneity. The appearing ofKinect sensor is a revolution for it’s cheap. This article make use Kinect to get the depthimage of body and model3D human body fast as well as real through the algorithm I propose.At the same time, in the field of graphics, more and more people put eyes on3D humananimation. In many applications, human animation is gained from specialized equipment and3D motion capture software (Maya, DAZ3D, Poser). The cycle for these software to getanimation is too long although motion capture animation data generated by the equipment hasmore accuracy and higher frequency. This is because the motion capture equipment iscomplex and wear-inconvenient. It’s difficult for human to perform with it. Therefore, thispaper is based on the existing skinning technology and manages to skin on the human bodymodel rapidly so that the result’s quality is improved.The article mainly discusses how to reconstruct a3D human body model fast in acomplex environment with a single Kinect sensor. There are five steps. Firstly, get the datastreams including Color, Depth and Skeleton in the complex scene. Then, analyze it anddivide human body’s data separately. Thirdly, mesh the3D human body data through theprocess on the cloud data of depth point which include the sub-process like denoising,smoothing, coupling and so on. What’s more, parameterize the standard model with thehuman body’s parameterized data. After fusing the body’s model and standard model, the goalthat optimize target3D human body model and generate real body fast can be achieved. Atlast, check the model interactively through using the real model generated by skeleton data.