复杂环境中多信息融合的手势识别.pdfVIP

180 2014 ，50（5） Computer Engineering and Applications 计算机工程与应用 复杂环境中多信息融合的手势识别 赵爱芳，裴 东，王全州，杨鸿武，安占福 ZHAO Aifang, PEI Dong, WANG Quanzhou, YANG Hongwu, AN Zhanfu 西北师范大学 物理与电子工程学院，兰州 730070 College of Physics and Electronic Engineering, Northwest Normal University, Lanzhou 730070, China ZHAO Aifang, PEI Dong, WANG Quanzhou, et al. Gesture recognition fused with multi-information in complex environment. Computer Engineering and Applications, 2014, 50（5）：180-184. Abstract ：Aiming at the problems existed in hand gestures recognition in complex environment, a hand gestures recogni- tion approach fused depth information with infrared information is proposed in this paper. Firstly, depth information captured by Kinect is applied to segment the dynamic hand gesture, then, using infrared image to repair the hand gesture image. It solves the problem of real-time gesture segmentation and low recognition rate of gesture recognition using distribution fea- ture of hand, when there is a defect or face interference in gesture segmentation region. The experimental results show that the proposed method is not sensitive to ambient light, and can identify gestures with small discrimination, and can recognize gesture of rotation, scaling, translation with strong robustness. For those gestures with large discrimination, the recognition rate is up to100%. Key words ：hand gesture segmentation; space eigenvector; hand gesture recognition; depth information; infrared informa- tion; complex background 摘 要：针对复杂环境中的手势识别问题，提出了一种融合深度信息和红外信息的手势识别方法。首先利用Kinect 摄像头的深度信息进行动态实时手势分割，然后融合红外图像复原手势区域。解决了实时手势分割和利用手势的 空间分布特征进行手势识别时由于分割的手势区域有缺损或有人脸干扰时识别率低的问题。经实验验证，提出的 方法不仅不受环境光线的影响，而且可以识别区分度较小的手势，对旋转、缩放、平移的手势识别也具有鲁棒性。对 于区分度较大的手势，识别率高达100% 。 关键词：手势分割；空间特征向量；手势识别；深度信息；红外信息；复杂环境 文献标志码：A 中图分类号：TP391 doi ：10.3778/j.issn.1002-8331.1304-0011 随着计算机技术和人工智能的发展，人机交互作为 前提下结合手势条件或者手势的几何特征[2-5] 。手势表 一项不可或缺的关键技术，已不再简单地局限于传统的