Registration and integration of multiple depth images using signed distance function

摘要

Depth camera is a new technology with the potential of radically changing the way to record and interact with 3D virtual environments. With depth cameras, one has access to depth information up to 30 frames per second, which is much faster than previous 3D scanners. This speed enables new applications where objects are no longer required to be static for 3D sensing. There is, however, a trade-off between the speed and the quality of the results. Depth images acquired with current depth cameras exhibit a high level of noise and have low resolution, which pose a real obstacle when incorporating the 3D information into computer vision techniques. To overcome these limitations, the speed of depth camera could be leveraged by combining data from multiple depth frames together. Thus, good registration and integration methods are needed that are specifically designed for such low quality data. Towards this goal, this paper presents a novel method for the registration and integration of multiple depth frames over time onto a global model represented by an implicit moving least square surface. Experimental results demonstrate promising high performance compared to existing methods.