3D Shape Recovery Based on Radon Transform
Chuchart Pintavirooj, Manas Sangworasil · 2001
ion The focus of this paper is to present a new approach for recovering 3D shape of an object. We exploited the idea of image reconstruction from projection to construct the cross-section image of the object. Each of the crosssection images is computed from a series of photographs taken at a number of angles around the object. A corresponding intensity profile (row) on each digitized photograph resembles a projection data, which can be used to reconstruct a cross-section image of the object. To extract a 2D coordinate of sectional contour of the object, edge detection is performed followed by locating the x-y position of the pixel constituting the detected edge. Stacking of 2D contours yields a 3D coordinate of the object. To provide better 3D visualization, we also performed a surface-rendering technique on the volumetric data constructed from a stack of cross-section image. Unlike conventional 3D-shape recovery method which uses stereo pairs of camera, our technique does not suffer from a laborious corresponding problem. The 3D – acquisition system is simple, as it requires only a digital camera and a rotating platform. Our purposed method is tested to perform 3D-shape recovery of a variety of objects. The result is very promising.