基于手势识别技术的3D虚拟交互系统  

A 3D Virtual Interactive System Based on Gesture Recognition Technology

作　　者：王粉花[1] 张万书[1] 

机构地区：[1]北京科技大学自动化学院,北京100083

出　　处：《实验室研究与探索》2014年第6期94-98,共5页Research and Exploration In Laboratory

摘　　要：系统以英特尔凌动处理器嵌入式平台为核心,运用手势识别、多目视觉立体定位、电机控制及3D建模等多种算法和技术,构建了一个真实人与虚拟世界进行实时交互的浸入式环境。一方面以图像处理算法为核心,进行基于肤色模糊识别和多智能体协作思想的图像分割,实现了基于形状特征的手势识别算法;另一方面采用了3dsmax辅助建模的方法,实现了手的骨骼模型这种复杂物体的建模,结合Direct3D开发3D虚拟环境系统,采用DirectSound产生环境的声音。通过对虚拟手的控制实现人与虚拟环境中三维物体的交互操作,如抓取、移动物体等动作。随着3D场景和动作的切换,伴随发出不同的声音。测试表明,该系统的手势识别成功率达到80%,人机交互较为生动、直观和真实。The core of the system is an embedded platform based on the Intel Atom Processor. Various algorithms and techniques, such as gesture recognition, stereo positioning based on muhi-camera vision, motor control and 3D modeling etc, have been employed to construct an immersing environment, in which a man can interact with the virtual world. On the one side, with image processing algorithms as the core, the image segmentation, which is based on skin color fuzzy recognition and multi-agent coordination, and a gesture recognition algorithm based on shape feature have been implemented. On the other side, by using 3dsmax ancillary method a complex object model, the hand skeleton model, have been established. Combined with Direct3D, a 3D virtual environment system was developed and the DirectSound was used to achieve 3D sound. Through the control of the virtual hand, man can perform interactive operation such as grabbing, moving the 3D object in the virtual word etc. With the updating of 3D scenes and switch of actions, different sounds can be heard. The test shows that the success ratio of gesture recognition of the system is up to 80% , what is more, human-computer interaction is lively, intuitionistie and real.

关 键 词：手势识别 虚拟交互 电机控制 3D建模 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]