一种人脸对象的区域分割方法

人脸特征的提取通常是先把人脸中包含特征信息的区域分割出来,然后分别对每个区域进行特征定位和提取。这样可以克服将整个人脸作为一个研究对象来处理所存在的复杂性,减少背景和其它物体的干扰,提高特征提取的准确性。还可以减少图像数据处理量,节省处理时间。在分析用在人脸对象区域分割的常用方法基础上,采用基于眼间带状人脸区域垂直梯度图的人脸对象区域分割方法对五官进行分割。实验结果表明:这种方法具有很高鲁棒性,并且算法实现简单。

A kind of segmentation method of face object

The first stept of face feature extraction is to devide out the regions which contanined characteristic information, then make characteristic position and extraction respectively of every area.This method can overcome the complexity of studing on the whole face,reduce the interference of background and other objects, improve the accuracy of characteristic extraction,besides,it can decrease image data processing and save time. Based on analyzing the common ways for regional devision,in this article, use the eyes of the zonal face regional vertical gradient to separate objects area of face teatures.The experimental results indicate that the method proposed here has good robustness, and the algorithm is easy to process.