Hierarchical Collision Detection Methods

简介:
In this section, we will discuss briefly the approach taken by hierarchical collision detection methods. These algorithms work on the face-face level. Given a pair of objects, they check which faces of the objects overlap, so they are carried out behind a method such as Sweep and Prune in the collision detection pipeline. Hierarchical Collision Detection algorithms approximate the objects in the scene using bounding volumes. Without specifying which type of bounding volume is used, the approach goes as follows. Given a set of polygons, calculate the bounding volume of this set. Next we construct a number of subsets that are maximally separated. For each of the subsets we calculate the bounding volume and link it to the parent node. This continues until the set contains some minimal number of polygons, usually just one.

At collision detection time, intersection is determined by first checking whether the bounding volumes at the roots of the trees corresponding to the objects intersect. If they do, we check the children of one of the nodes against the other node. If the two nodes being tested are leaves, the faces contained in the leaves are checked against each other and added to a collision list if they intersect. Like this, we continue until no more intersections can be found.

In this implementation, we used Oriented Bounding Boxes (OBB's) and Axis aligned Bounding Boxes (AABB's) as bounding volumes, since they are easy and fast to construct and for both types of objects rapid overlap tests exist ([GLM96],[Ber98]). With the OBB's we have a further option of calculating the convex hull when computing the orientation. Of course, many other types of bounding volumes exist like spheres and k-DOP's etc. I won't go into this here, but rather move on to the implementation of the parallel versions of the OBB and AABB based algorithms.

目录
相关文章
|
运维 算法 数据挖掘
Outlier Detection Methods|学习笔记
快速学习 Outlier Detection Methods
Outlier Detection Methods|学习笔记
|
算法 数据可视化 数据挖掘
Hierarchical Methods|学习笔记
快速学习 Hierarchical Methods
Hierarchical Methods|学习笔记
|
索引
Mark Graham - Cloud-based Detection Techniques for Botnets and Other Malware
http://www.youtube.com/watch?v=fV5kED7nryw&list=PLpr-xdpM8wG_KHsxepT9o6trkqDELhr3_&index=12 ...
744 0
|
资源调度 算法
论文阅读之 DECOLOR: Moving Object Detection by Detecting Contiguous Outliers in the Low-Rank Representation
DECOLOR: Moving Object Detection by Detecting Contiguous Outliers in the Low-Rank Representation Xiaowei Zhou et al.
|
存储 机器学习/深度学习 人工智能
PTPCG: Efficient Document-level Event Extraction via Pseudo-Trigger-aware Pruned Complete Graph论文解读
据我们所知,我们目前的方法是第一项研究在DEE中使用某些论元作为伪触发词的效果的工作,我们设计了一个指标来帮助自动选择一组伪触发词。此外,这种度量也可用于度量DEE中带标注触发词的质量。
145 1
|
机器学习/深度学习 移动开发 算法
DISCOBOX: Weakly Supervised Instance Segmentation and Semantic Correspondence from Box Supervision
定位和识别物体的能力是人类视觉的核心。这促使视觉社区研究对象检测 [1] 作为一项基本的视觉识别任务。在检测之上进一步引入实例分割 [2] 以预测前景对象掩码,从而实现像素级精度的定位。
108 0

热门文章

最新文章