基于HSV空间优化耦合参数PCNN的人眼定位方法

Eye Location Based on HSV Space Optimized Coupling Parameters of PCNN

为解决人眼定位中受光照、人脸姿态、人眼开闭影响的问题,提出了一种HSV空间肤色模型与优化耦合参数的脉冲耦合神经网络(Pulse Coupled Neural Network,PCNN)相结合的人眼定位算法。将RGB空间图像转化为HSV空间图像,根据肤色检测人脸图像,利用优化耦合参数PCNN模型良好的捕捉特性及衰减特性,对图像进行分割,提取出人眼范围,利用形态学运算对图像进行增强处理,去除背景干扰,精确定位人眼。实验结果表明,所提算法在不同光照条件、不同人脸姿态,人眼闭合均能成功定位,且具有定位准确、速度快、不受人眼开闭程度影响的特点。

The paper proposes an algorithm of human eye location, which combines skin color model with optimized coupling pa- rameters PCNN model. Transforming RGB space image into HSV space image, detect skin color. Then detecting the image of hu- man face based on skin color model. In order to separate out the scope of the human eye, properties of coupling parameters opti- mized PCNN model of well capturing and attenuation are employed to segment the image. Conducting image enhancement by com- bining tile top hat transformation with a low cap transformation. Implementing morphological open operation processing on the en- hanced image, removing background noise, and loeking the accurate positioning of the eyes with the gray curve. According to the result of simulation, it shows that this algorithm can achieve accurate positioning of human eye under differently illuminant background. This algorithm has superiorities both in positioning accuracy and high speed. Moreover, it is not affected by human eyes＇ states, which means it can operates well in all conditions no matter the human eyes are opening or closing.