眼部动作实时识别系统设计

Design of Real-Time Blink Detection System

利用网络摄像机和图像处理单元设计了一种眼部动作实时识别系统.该系统首先用帧差法检测出眨眼的动作区域,将此区域设置成感兴趣区域,并用图像形态学算法对其进行处理;然后找出眼部轮廓,并以此区域创建模板,再利用此模板对每一帧图像进行模板匹配,找出匹配中的最小值,以得到识别的结果;最后通过实验验证了该算法的有效性.该系统有广泛的应用前景,如实现模拟鼠标的控制等.

A system based on network camera and image processing unit was designed to recognize eye movement real-time.Firstly,the region of blink,which is confirmed by difference in frame difference,is set as regions of interest（ROI）,and then ROI is processed by morphological operation.Secondly,the eye contour is extracted to create a template region,which is used to mask image,and this template is used to match each frame in order to find the minimum to obtain recognition results.Finally,the validity of the algorithm was proved by experiment.The system has a wide range of applications such as control of the mouse,etc.