几何性特征的眉毛模型的研究  

Study on Eyebrow Model Based on Geometric Features

作　　者：梁奔香 蔡晓东[1] 朱利伟[1] 华娜[1] 吴迪[1] 

机构地区：[1]桂林电子科技大学信息与通信学院,广西桂林541004

出　　处：《电视技术》2015年第13期45-49,共5页Video Engineering

基　　金：国家科技支撑计划课题项目(2014BAK11B02);广西省自然科学基金项目(面上项目)(2013GXNSFAA019326);桂林电子科技大学研究生科研创新项目(GDYCSZ201410)

摘　　要：提出一种新的基于几何性特征的眉毛模型,相比传统的基于眉毛的全局数学特征建立的眉毛模型,它可以直观地描述和反映眉毛的几何外观形状,这种特性在人脸识别等基于面部视觉特征的应用是关键的。首先获取人脸单边眉毛图像并统一归一化尺寸为80×30的大小,使用阈值化和Blob分析得到眉毛归一化的二值图像,然后由二值图像计算眉毛模型的几何参数,包括眉毛区域的面积、长度、平均宽度、弯曲程度和走向趋势等。实验表明,提出的眉毛模型可以准确有效地描述眉毛的几何外观形状。A novel eyebrow model is proposed based on geometrical characteristics of eyebrows. Compared to the traditional eyebrow model that based on the global mathematical features, it can describe and reflect the eyebrows geometric shape intuitively. This characteristic in application such as face recognition based on facial features is critical. Firstly, single side eyebrow image should be got, and normalized the size with 80 × 30, then using the threshold and Blob Analysis to get the binary image of brow. Secondly, the geometric parameters of the eyebrow model are calculated, including the area of eyebrow, the length, the average width, the trend of the bending degree and the orientation. Experimental results show that the eyebrow model of this paper can describe the eyebrow shape accurately and effectively.

关 键 词：眉毛 几何特征 眉毛模型 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]