基于约束局部模型的人眼区域检测研究

Research on Eye Region Detection of Based on Constrained Local Model

为了提升人眼区域检测的精确性与稳定性,通过基于统计学习的逻辑回归法和卷积神经网络法2种不同的局部检测器,提出一种基于约束局部模型的人眼区域定位方法。首先,在原始图片中检测出人脸区域;然后,在人脸区域中定位出人眼关键点位置,将关键点的外界矩形作为人眼区域的检测结果。结果表明:基于约束局部模型的人眼区域检测方法不仅对不同肤色、光照及头部姿态的测试者都能精确检测出眼睛区域,基于卷积神经网络的局部检测器的检测结果也更加准确。基于约束局部模型的人眼区域检测方法具有较好的精确性与稳定性。

In order to improve the accuracy and stability of human eye region detection,this paper proposes a human eye region localization method based on constrained local model through two different local detectors:logical regression method based on statistical learning and convolution neural network method.Firstly,the face region is detected in the original image,and then the location of the key points in the face region is located,and the outer rectangle of the key points is taken as the detection result of the human eye region.The results show that the eye region detection method based on constrained local model can accurately detect the eye region of volunteers with different skin color,illumination and head posture,especially the detection result of local detector based on convolution neural network is more accurate.It can be seen that the human eye region detection method based on constrained local model has good accuracy and stability.