基于人眼注视非穿戴自然人机交互

Gazing Based Non-Wearable and Natural Human-Computer Interaction

提出了一种基于人眼注视的非穿戴自然人机交互新方法。基于人体生物结构特征,采用主动形状模型确定人眼轮廓特征点,并根据HSV色彩空间构建人眼特征直方图,采用粒子滤波法,对人眼目标跟踪与定位。基于最大三角化划分人眼轮廓特征,构建人眼几何模型,通过图像帧间均值滤波,确定人眼注视交互目标,实现非穿戴的人机交互,满足用户交互的灵活性、舒适性和自由性等要求。通过实验对比,验证了该方法有效、可行。

A novel non-wearable and natural human-computer interaction（HCI）method has been proposed based on eye gazing. According to human being biological structure characteristics ,an active shape model is employed to locate some feature points in the eye profile. A histogram of eye feature has been built according to the HSV color space. A particle filter method has been adopted to track and locate the eye. A 2D eye geometric model is constructed based on the maximal triangulation of the eye contour features. A temporal median filter strategy has been developed to determine a stable gazing interactive target. Non-wearable and natural HCI modal is realized in which the user can move flexibly both in comfort and freedom interactive ways. Experiment results indicate that the developed approach is efficient and can be used to natural non-wearable HCI.