Non-photorealistic Rendering Based On The Point Cloud Technology Research

With the rapid development of 3D scanning technology, Point cloud has become one of the most popular 3D modeling techniques. For point cloud models, the data is easier to acquire and its data structure is much simpler than traditional mesh models. With these advantages, point cloud is very good at representing extremely large or complex models, such as sculpture and other irregular shaped objects.The surface information of a point cloud model is very abundant. It can include: position, normal, size, color, texture and transparency etc. Point cloud surface is the real meaning free-surface, and can represent arbitrary shape.The research of point cloud has become a hot spot in recent years.Point cloud is direct sampling points of the object surfaces by sampling objects using the three-dimensional scanner, and is the most real record of the object's surface geometry attributes. Currently, data processing of based on point cloud model mainly includes: data compression, data transmission, geometric modeling, feature detection and rendering system etc.Silhouettes play an important role in the non-photorealistic rendering for 3D shape, and they are necessary for both perception and aesthetic, which represent the contour and basic shape of the 3D models. Therefore, artists and designers often emphasize the silhouettes: either drawing them explicitly or by contrast enhancement across silhouette boundaries. In many hand-drawn 3D scenes, in order to highlight the designers'design intent, we often show the important characteristics of model through simplifying some of the local details which are not very concerned about, but thickening the contour lines. So, silhouette line extraction technology holds an important position in the non-photorealistic rendering.Point clouds are discrete data points without the topology relationship between them. So, to facilitate and expedite the extraction of contour points for the late NPR style, firstly we should simplify the point cloud data. In other words, non-contour points should be eliminated but their contour points and feature points should be retained .Secondly, we apply the criteria to extract contour points and form the contour lines. Finally, the contour lines can be processed for the NPR effects. That is the focus of this paper and is the processing process for the point cloud data.This paper is about non-photorealistic rendering technology based on the point cloud data, and in the rendering process, we make the full use of the contour lines'important role in the NPR. This paper's main contributions in the following are as:1.Point cloud data's reduction:For the effects of NPR ,we do it by the help of the contour lines ,which are from the contour points. To obtain contour points, we first"roughly"simplify the point cloud data. We presented the method of simplifying the point cloud data, in which we remove non-feature contour points of the object using the change of normal vector between the points.2.Extraction and mergment of the contour points for point cloud data:we extract contour points of the object from the point cloud data by use of the object's curvature. What is more, for the effective process of NPR style, contour points should be combined to silhouettes under the "direction" of the contour points.3.The process of NPR style effects:During the NPR effect process for point cloud data, the "encryption" handling is put to the contour points, which completes the silhouette's increase of the object contour,strengthens the boundary contrast of the object shape, achieves good visual effects, and creates a sense of beauty.4.Experimental platform environment and the process: The hardware experimental platform environment mainly includes: Intel Pentium E2140 dual-core processor, two Kingston 1G DDR2 memory etc. The algorithms were validated and tested to obtain the desired results, using Visual C++ 6.0 and OpenGL.