3D Image Understanding and Recognition in Virtual Environment.pdf

3D Image Understanding and Recognition in VirtualEnvironmentPatrick S. Wang, IAPR FellowCollege of Computer ScienceNortheastern UniversityBoston, MA 02115e-mail: pwang@/home/pwangAbstractThis article discusses the state-of-the-art ideas in high level visualization, understanding andinterpretation of line-drawings images. A new strategy using parallel generalized linear combinationmethod is presented. It is simple and needs only very few learning samples. Further, it can strengthenadvantages of current key methods while overcome their drawbacks. Also, it is able to distinguish objectswith very similar patterns and is more accurate than other existing methods in the literature. In addition,it is readily applicable in virtual environments including internet, mulit-media, and WWW hompeageimplementations for image understanding, visulaization and interpretation.1 IntroductionIt is well known \One picture is worth one thousand words. Indeed, it can be easily seen frommountains of documents, among other things, many of which include images, scenes, graphics,photos, or pictures. They are useful to illustrate certain ideas, to demonstrate some meth-ods and algorithms, to describe a situation or scene, or to interpret a CAM/CAD engineeringdesign. Therefore, how to understand these images attracts increasing attention among re-searchers, engineers, scientists, government administrators, and industrialists in recent years.[7,11,19,20,35,43,46]Line drawing images, also called \noble class of images by Sugihara [33] in which manyimportant and essential phenomena of 3-D objects hold in the real world, provide a very eectiveand practical method to describe the 3d shape of an object. These drawings normally can beorganized into two or more distinct views, known as characteristic views or aspects in which noone view by itself is sucient to completely characterize the object being represented.(Lysak andKasturi)[21] One of the most challenging and dicult problems we will face in doc