Fast Selecting Visible Surfaces in High Precision
快速高精度的可见面选择

It is very important to efficiently select visible surfaces in rendering complex scenes. As for this, a method is presented in this paper by combining the strategies for image-based selection and object-based selection. First, it classifies the surfaces by their normals, and manages every class of surfaces in hierarchical indexing structures by their positions in 3D space. Then, the visible surfaces are pixel-driven searched from near to far. As the index structures can manage the surfaces orderly, this search can be calculated quickly and all visible surfaces can be obtained. Meanwhile, the selected visible surfaces can be fused automatically to act as the occluders to implicitly cull the occluded surfaces in large quantities. Compared with existing methods for selecting visible surfaces, the new method is fast and in high precision, and can conveniently handle the scenes with dynamic objects.