人眼定位方法的研究目前已经成为人脸识别智能化中的重要研究内容。提出了一种无需先确定人脸位置,而在有背景灰度人脸图像中直接检测与定位人眼的新算法。利用复杂度的原理,对人脸图像中眼睛位置进行快速定位,剪切眼睛区域的图像块;再利用图像分割的原理,自动寻找最佳闽值,提取眼睛的边缘;进行形态学操作,排除非人眼的图像块,从而正确定位眼睛的位置。该方法能较大提高人眼定位的计算速度,适用于实时性要求较高的场合。仿真实验验证了该算法的有效性和稳定性。
The eyes location problem in intelligence of human face detection is studied. A new 'algorithm that can detect directly and orient the human eyes in human face images with background gradation is proposed, and it does not need to determine the human face position firstly.The method uses the conception of the complexity extent to carry on the fast orientation for the eyes position in human face images, and it cuts the image blocks near eyes. Then it uses image division principle to seek the best threshold value automatically and withdraws the edges of the eyes. It finally carries on the morphology operation In remove the image blocks of the human eyes, thus orienting the position of the eyes correctly. Massive simulation and experiments show that the proposed method is effective in raising the speed of eyes location, and can be applied to realtime situation.