Local Shape Similarity Measure Of3D Meshes And Its Applications

The3D mesh model is widespread used in computer to display objects in computer graphics,3D mesh can be obtained through3D scanner or some3D modeling software. With the rapid development of3D model technology, in order to show more reality and using discrete poits in computer to form the continuous physical nature, we need to do some analysis and remeshing for the preliminary3D model. Local shape similarity is widely used in computer graphics and computer vision.The main work and achievements in this paper are as follows:1)We presented a novel local shape similarity measue which is based on spin image in this paper. For efficiently comparing two different regions on3D mesh model, the neighboring vertices for each surface vertex are firstly up-sampled by uniformly sampling the tangent plane and re-projecting onto the3D surface. The densely sampling vertices are then distributed uniformly along the geodesic lines on the local region of the underlying surfaces. The Spin images are thus obtained for defining subsequent local shape similarity measure.2)Due to our proposed local shape similarity definition, an efficient visual enhancement scheme is provided by transporting the user defined vertex color onto similar local regions. And then we applied the local shape similarity measue to3D model denoising when combine Non-Local means for2D image.Experimental results illustrate that our spin image based local shape similartity definition is robust to the application of similarity guided visual enhancenment. In addition, the combination of local similarity measure and non-local means for2D iamge is very efficient for maintaining the inherent feature of model in3D model denoising.