school
Macau Periodical Index (澳門期刊論文索引)
- Author
- Wong, Unhong; Wong, Honcheng; Tang, Zesheng; Tian, Xiaolin
- Title
- Point-based representation of the segmented visible human dataset
- Journal Name
- 澳門科技大學學報
- Pub. Info
- Dec. 2007, Vol.1, No.2, pp. 22-32
- Keyword
- point-based representation；volume visualization；segmented Visible Human dataset
- Abstract
- Abstract : Triangle mesh is still be most common surface representation in many computer graphics applications. Because of its simplicity and flexibility, it is widely used in many areas where processing performance matters. On the other hand, point-based techniques are better in 3D modeling while point-based representation is good for efficient rendering and flexible geometry processing of highly complex 3D models. Based on our knowledge and survey of point-based representation, we consider that point-based representation is suitable for representing the segmented visible Human dataset. In this paper, a point-based representation for interactive visualization is proposed, and it is applied to the segmented Visible Human dataset, from building the 3D models, specifying normal vectors, smoothing and performing some other enhancements, are discusse. Furthermore, a hybrid representation method for volume visualization has also been presented, implemented and applied to modeling and rendering of the segmented Visible Human dataset, trying to improve the performance. The idea of hybrid representation is to use both points and meshes(triangles, quads, or polygons) for modeling and representing 3D models. Deformation and internal viewing features are provided as well as the internal structure of the human body can be displayed with our approach. Experimental results show that our approach has improved both quality of the model and rendering speed. The rendering speed (around 20 to 30 FPS) is fast enough for most interactive applications while previous approach using marching cubes to create 3D models of the segmentaed Visible Human dataset only gets about 10 FPS. Paragraph Headings： 1. Introduction 2. Related work 2.1. Segmented visible human dataset 2.3. Interactive volume visualization of Segmented Visible Human Dataset 3. Our approach 3.1. Extraction of the point-sampled surface from the dataset 3.2. Specification of Normals 3.3. Smoothing of the Models 3.4. Hole filling method 3.5. Backface culling method for surfels 3.6. Antialiasing 4. Hybrid 3D models of the segmented visible human dataset 4.1. Finding and reconstructing surfaces 4.2. Rendering of hybrid models 5. Internet structure visualization 6. Deformation 7. Result and analysis 8. Conclusion 9. Future Work Figures: 1. Overall working pipeline 2. Specification of normals 3. Calculations for smoothing 4. Model without and with smoothing 5. Memory arrangement scheme for smoothing 6. Back-face culling method for surfels 7. Filtering: calculate the weighted average with neighboring pixels 8. Image without and with antialiasing 9. Extracting flat surface of the point-sampled object 10. Hybrid representation of the segmented visible human dataset 11. Viewing the internal structure 12. Bump-mapping concept 13. Deformation results 14. Rendering of the whole body with segmented visible human dataset 15. Rendering 3D models with internal viewing