基于PCNN的人脸图像分割研究

Research of Face Image Segmentation Based on PCNN Algorithms

本文深入研究脉冲耦合神经网络的基本原理,对原始PCNN模型做了一定程度的简化。利用简化型PCNN对人脸图像进行分割,性能优于其它算法。该方法完全依赖于人脸图像的自然属性,不用预先选择处理的空间范围,利用人脸图像的灰度特征和空间信息的相关性,得到了人脸图像的神经元点火序列,该点火序列就是图像分割的结果。实验结果证实了本文算法的有效性。

This thesis investigates the model of PCNN,which includes its basic principles. Through reducing some parameters we get a simplified PCNN.Simplified PCNN can be used in image segmentation,which is Superior to other algorithm.This algorithm entirely relies on the nature attributes of facial image,which doesn't have to select the scope of space. It utilizes the gray feature of image and relativity of spatial information,so the algorithm gets the nerve cell ignition sequence of image which becomes the result of image segmentation. The experiment results demonstrate that the algorithm is valid.