人机交互中手势图像手指指尖识别方法仿真

摘要

A fingertip recognition method of gesture image in the human-computer interaction is proposed based on the depth information.Firstly,the Kinect is used to acquire the depth image of palm and the image is separated from the background.Then,the two-dimension outline of each finger integrated with the morphological operation in the binary image of palm is worked out according to the binary image obtained by the mapping.Moreover,the point farthest from the palm center point in each finger outline is defined as the fingertip point,and the curvature method is used to exclude the crooking finger and finger groove.Finally,the fingertip recognition in the human-computer interaction is completed.The simulation results show that the method has high recognition precision.It can recognize the track of finger tracing accurately.%人机交互中对于手势图像指尖识别方法的研究,能够使之更好的适应各类人群,减小传统人机交互的局限性.对于人机交互中手势的指尖识别,需要计算出各个手指的二维轮廓,利用曲率方法排除弯曲的手指和手指凹槽.传统方法需要计算出摄像机的标定参数,利用该参数得到两个图像中的指尖位置,但没有排除手指弯曲、凹槽部位,导致识别结果不准确.提出基于深度信息的人机交互中手指指尖识别方法.利用Kinect获取手掌的深度图像,将图像从背景中分离,依据映射获得手掌的二值图像,融合形态学操作在手掌的二值图像中计算出各个手指的二维轮廓,将每个手指轮廓中距离掌心点最远的点定义为指尖点,利用曲率方法排除弯曲的手指和手指凹槽,从而完成对人机交互中指尖识别.仿真结果表明,所提方法识别精确度高,可以准确识别出指尖跟踪的轨迹.