几何性特征的眉毛模型的研究

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.

Study on Eyebrow Model Based on Geometric Features

This paper proposes a novel eyebrow model 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 that 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 were 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.