虚拟环境下的多维视觉几何匹配方法研究

Research on Multi-Dimensional Visual Geometric Matching Method in Virtual Environment

针对传统虚拟装配系统存在人机交互视角单一,操作者因视觉信息不足致使训练效率不高、极易产生训练疲劳的问题,提出了多维视觉几何匹配方法。该方法根据装配对象的位置信息,通过几何法引入多个视觉点,解决装配过程中的视角信息反馈不足问题。为验证该方法的有效性,在Visual Studio 2010开发环境下,搭建了基于Geomagic Touch力反馈设备的虚拟装配系统。将力反馈设备作为人机交互实体装置,由3DSMAX绘制虚拟装配系统中的目标物体,经由Open GL进行渲染并显示。试验结果表明,与传统单一视角的装配系统相比,该方法缩短了35%以上的操作者训练时间,有效解决了视觉信息反馈不足的问题,提高了操作者在装配中的训练效率,同时优化了人机交互性能。

Aiming at the disadvantages of traditional virtual assembly system, e. g. , unitary angle of human computer interaction, low training efficiency of operators due to weak visual perception information which leads to training fatigue, a multiple dimensional geometric matching method is put forward. According to the position information of the object assembled, through the geometric method, multiple visual points in the vicinity of the target are introduced, to increase operator perspective for solving the problem of insufficient feedback from perspective information. For verifying the effectiveness of this method, under Visual Studio 2010 development environment, the virtual assembly system is built by force feedback device named Geomagic Touch. In the system, the force feedback device is used as the human-computer interaction entity device. The target object in the virtual assembly system is drawn by 3DSMAX, and it is rendered and displayed by OpenGL. The virtual assembly simulation experiments under single perspective and multiple perspectives are completed by multiple users. The experimental results show that compared with the traditional single perspective system, this method can solve the problem about insufficient feedback of visual information, improve the training efficiency of the operators, and the training time is shortened about more than 35 % , the human computer interaction performance is also optimized.