Estimation of Principal Component on 3D Objects.

Abstract

It is proved that the direction of lines exerts a significant influence on human‟s perception of object shape. Furthermore, the lines in the principal direction of curvature on an object rather than an arbitrarily selected one can provide very important details in depicting object shape and silhouette edges. The major goal of this project is to effectively estimate a collection of principal direction vectors on a computer generated 3D object. On the basis of these vectors, basic line drawings can be performed in order to reveal how these lines in the principal direction delicately reflect the object shape. Throughout this project, the concept of principal direction and its related applications in representing 3D object have been carefully studied. Major effort has been put into reproducing the process of estimating the principal directions based on a normal-ray focal surface model proposed by Jingyi Yu et al. in 2007. A collection of principal direction vectors has been successfully generated and presented onto the surface of each processed 3D object. In the end, extra focus has been placed on some current applications that aim at creating sketching effects on 3D models on the basis of their estimated principal components. Two line drawing techniques have been investigated in the process. Future work of this project is proposed to be concentrated on this field.