Research On Best Viewpoints Selection Of3D Models Based On Feature Extraction

With the development of three-dimensional modeling techniques andthe innovation of computer network technology, it becomes easier andeasier to obtain3D models. The widely applications of3D modelsmakes recognition of3D models be an important applicationrequirement. Therefore, a technology named viewpoint selection whichmakes computer enable to select representative observation viewpointsfrom3D space quickly and automatically then arises at the historicmoment. Viewpoints selection of3D models is a fusion technology ofthe computer intelligence, image processing, computer vision, and so on.Best viewpoints selection can choice high quality viewpointsintelligently as well as quickly. It has broken through the limitations oftraditional manual selection, and has become a hot research issuerecently.Information extraction and measurement of3D surfaces are twocritical issues involved in the process of viewpoints selection. This paperproposes best viewpoints selection of3D model based on a featureextraction process, and introduces the corresponding viewpoint qualitycalculation standard based on the extracted feature. A3D model bestviewpoints selection system is then presented in the final. Main contents and works of this paper are shown below:1. Viewpoint-dependent statistical feature extraction of3D surface.For a3D mesh model, inspired by the D1distance of shapedistribution, we propose a viewpoint-dependent statistics featureextraction method. In comparison with the related statisticalinformation extraction methods, our approach is not only consideredthe three-dimensional surface, but also take into account the positionrelationship between the viewpoint and the three-dimensional model.2. Viewpoint quality calculation based on the viewpoint-dependentsurface statistics features. For a candidate viewpoint, compute thedistances between all visible sampled points and that viewpoint, thenhistogram of the distances. Inspired by the ideas of the literature [2],the Shannon entropy of the histogram is regarded as the quality ofthe candidate viewpoint.3. Feature points detection of3D surfaces based on a novel vertexsaliency measurement. We assumes that the saliency of a givenvertex on a3D mesh model can be described by its refinement andits height within a local space, and then the effective feature pointsare promisingly able to be extracted combined with extremesuppression.4. Viewpoint quality measurement based on the surface feature points. The quality of viewpoints is a combination of geometric distributionand saliencies of all visible feature points. In comparison with therelated works, this viewpoint quality calculation standard depends onthe visible feature points only.5. A best viewpoints selection system for3D models is designed.Comprehensive our research works in this paper, we design andimplement a viewpoint selection system of3D model. Methodsabout viewpoint quality measuring proposed in this paper are appliedand validated in this system. In correlative projects, the system canplay a role for some applications.