人机交互过程中数字手势的识别方法

为了获得简单、高效的数字手势识别方法,增加使用者舒适的体验,提出一种基于Kinect融合深度信息和骨骼信息的数字手势识别方案.首先,使用Kinect进行深度数据的采集,建立深度图像;其次,结合骨骼追踪系统,提取人体轮廓,运用深度阈值法从轮廓中分割出手部区域,并进行二维图像的重建;再次,利用手腕和手掌骨骼点准确分割出手掌区域,并运用图像形态学开运算进行处理,得到不含手指的图像,进而提取掌心坐标;最后,计算半径,确定掌心圆,采用圆的边界和手指相交次数的方式识别手指个数.实验结果表明:数字手势识别方案能够准确、高效地识别数字手势.

Recognition Method of Digital Gestures in Human-Computer Interaction

To obtain a simple and efficient digital gesture recognition method and increase the user’s comfortable experience, a digital gesture recognition scheme based on Kinect fusion depth and bone information is proposed. Firstly, the scheme uses Kinect to collect depth data and establish depth image. Secondly, the bone tracking system is used to extract the contour of the human body. The depth threshold method is used to segment the hand region from the contour and reconstructing the two-dimensional image. Next, the skeleton points of wrist and palm are used to segment the palm area accurately, and use morphological opening operation to obtain the image without the finger, further extract the palm coordinates. Finally, the radius is calculated to determine the palm circle, the number of fingers was judged by the times the circle boundary intersected with the fingers. The experimental results show that the digital gesture recognition scheme based on Kinect fusion depth and bone information can recognize digital gestures accurately and efficiently.