基于视点的三维点云自适应多细节层次模型动态绘制

在计算机几何图像设计、医疗研究诊断、物体判定识别、自动化测量等领域应用的迫切需求下,需要对三维点云动态绘制方法进行研究。目前存在的基于图形处理单元(GPU)的三维点云动态绘制方法。首先根据高度场的形式获得三维点云数据;然后利用高通滤波法对获得的三维点云进行消噪处理;再根据Scene Graph的方法表示三维点云之间的逻辑关系;最后利用GPU完成三维点云的动态绘制。为了降低噪声对三维点云绘制的影响,绘制更加流畅清晰的画面,提出一种基于视点的三维点云自适应多细节层次(LOD)模型动态绘制方法。首先利用激光扫描仪器或结构光扫描仪器等获取三维点云数据,并利用双边滤波器进行平滑去噪;然后根据特征配对的方法对三维点云进行特征对的提取与配准;最后以三维点云特征对配准结果为依据,利用三维点云的内外存调度完成三维点云的自适应LOD动态绘制。实验结果分析证明,所提方法能有效地去除噪声,可以更好地描述目标物体或背景,使绘制画面更加流畅。

Based on a viewpoint of dynamic adaptive LOD rendering 3 d point cloud

In computer graphic design geometry, medical research areas such as diagnosis, judgement of object recognition, automatic measurement applications under the urgent needs of need for 3 d point cloud dynamic mapping method are studied. The existing 3 d point cloud based on GPU dynamic mapping method, first of all, according to the height field in the form of three-dimensional point cloud data obtained; Then high-pass filtering method is used to get for 3 d point cloud denoising; And then based on the method of SceneGraph said the logical relationship between 3 d point cloud; Finally using the GPU to complete the dynamic drawing of 3 d point cloud. In order to reduce the influence of noise to draw 3 d point cloud, draw more fluent, clear images, put forward a viewpoint based on the 3 d point cloud adaptive LOD dynamic mapping method, first using laser scanning instrument or instruments such as structured light scanning, and obtain three-dimensional point cloud data, and use the bilateral filter to smooth denoising; Then according to the characteristics of the matching method of 3 d point clouds for feature extraction and registration; Finally, three-dimensional point cloud characteristics as the basis, with the result of registration using 3 d point clouds inside and outside storage scheduling completed adaptive LOD dynamic rendering 3 d point cloud. Analysis of experimental results show that the proposed method can effectively remove the noise, can better describe the target object or background, drawing picture more fluent.