增强现实系统中基于先验知识的虚拟抓取识别

摘要

为了实现增强现实系统中抓取动作的识别,针对传统方法实时性不强、容易产生误判的不足,提出了基于先验知识的虚拟抓取识别算法.通过对大量实验数据的分析,提取关键数据进行处理并完成决策树的构建,通过基于一种数据拟合的碰撞检测方法,真正实现了真实手对虚拟对象的抓取识别.实验结果表明:该算法识别率高、实时性好、交互自然真实,较好地满足了增强现实系统中抓取交互的需求.%In order to solve the real-time problem and the misjudgment of the traditional method, the virtual grasping recognition algorithm based on prior knowledge is introduced in an augmented reality ( AR). Through analyzing a large number of experiment datas, the key datas are extracted and processed, the decision tree is constructed. A new kind of collision detection method based on data-fitting have been use to realize the recognition of grasping. The experiment results show that the algorithm has high recognition rate, well real-time, and the process of interaction is natural, which satisfies the requirement of interaction in AR system.