使用眼睛定位的头部姿态识别方法

head posture recognition method using eye location

  • 摘要: 本文提出了一种基于眼睛定位的头部姿态识别方法。该方法首先利用人脸区域中眉毛和眼睛灰度梯度变化大的特点来定位眉毛与眼睛区域;然后利用Hough圆检测方法检测瞳孔位置;在此基础上,对瞳孔位置进行修正;最后,利用人眼定位的结果来识别头部姿态。本文将该方法用于模拟鼠标操作。与已有眼球操控鼠标不同,本文方法是利用眼睛定位的结果来识别头部姿态,并用头部姿态来模拟鼠标操作。实验结果表明,本文方法对5种头部姿势的识别率达到95%。

     

    Abstract: This paper presented a novel head posture recognition method based on eye location. Firstly, the method took advantage of feature that eyebrow and eye shades of gray in the face region relatively large changes to locate eyebrow and eye area, and then using Hough circle detection detects pupil circle, followed by the pupil correction, finally, using the result of eye location to recognize head posture. This paper applied it to simulate mouse operation. Unlike previous simulated mouse, this article made use of eye location to identify head posture, and then emulated mouse operation. The experimental results show that making use of presented method, the recognition rate of five head postures is to 95%.

     

/

返回文章
返回