3D image reconstruction of single camera body scanner

The main function of a body scanner is to digitize an object for further processing in the virtual space. To reconstruct the 3D images, typically, three or more views from different angles are desirable. Correspondingly, three or more cameras or sensors are needed so that the object remains stationary. Unless the number of cameras or sensors is reduced, the dimensions of the scanner cannot be shrunk. When there is only one camera available, the object must turns before multiple images can be captured.||Theoretically, three views are sufficient to provide enough information for the reconstruction of 3D images, if the subject maintains relatively stationary in the images. If two views are used, there will be areas that cannot be captured. In such case, extrapolation is indispensable. Regardless the number of views, reconstruction of images can be handled by one of the following methods: (1) averaging, (2) seam matching, (3) derivative matching, (4) normal matching. In this article, we shall introduce how these methods can be applied in a single camera body scanner to 3D images. Then we shall discuss the key assumptions and limitations of each method, with a comparison