2D 和 3D IMAGE SKELETONIZATION.pptVIP

University of Houston Clear LakeComputer Applications UHCL Conference 2D 3D Image Compression with Skeletonization 2D 3D Image Compression with Skeletonization Abstract Skeletonization helps an object image easier to be read/used in object recognition. Skeletonization is also an important compression tool. Binary images is compressed/skeletonized, transmitted, and then reconstructed/decompressed at the destination. Animation uses skeletonization to perform movements first and then reconstructs the object. The demands of skeletonizaton appear in many areas. However, existing skeletonization techniques suffer connectivity loss, and are mostly sensitive to noise, very time consuming, or restricted to specific 3D models. This presentation shows an approach to overcome the weaknesses above as well as some practical results applied on rocks to help scientists evaluate oil/water/gas stream through the rocks. Acknowledgement: Partially sponsored by TX HECB ARP/ATP Index Introduction Previous works Our approach Implementation Some practical results Limitations Conclusion References I. Introduction Definition: The skeleton is an object reduction. Properties: Geometrically centered within the object boundary . Having the same connectivity with the object. Topology remain constant, and one pixel of thinning. 2D 3D Image Compression with Skeletonization II. Previous works – Boundary peeling/Erosion Description Operate on the boundary from outside to inside. Test each boundary voxel to see if it is removable. Peel off layer by layer. Advantages Connectivity preservation. Disadvantages Time consume Noise sensitive Hardly to extend to 3D 2D 3D Image Compression with Skeletonization II. Previous works – Distance Transformation based on Voronoi Diagram Description Based on Voronoi diagram concept. Suitable for polygonal defined object Advantages Accuracy and good in connectivity preservation. Disadvantages Complicated and slow Noise sensitive Hardly to extend to 3D