基于猕猴面部图像特征的计算机自动定位方法  

An Automatic Computer Method for Macaque Facial Features＇ Localization

作　　者：杜希扬[1] 张惠云[1] 

机构地区：[1]山东中医药大学中医药经典理论教育部重点实验室,济南250355

出　　处：《实验动物与比较医学》2011年第3期166-171,共6页Laboratory Animal and Comparative Medicine

基　　金：国家重点基础研究发展计划（973计划）资助课题（从肝论治抑郁症、经前期综合征探讨肝主疏泄调畅情志的作用机制,2011CB505102）;国家自然科学基金重点项目（肝疏泄失常与调肝方药对不同脑区功能的影响及其机制研究,30930110）

摘　　要：目的提出一种基于猕猴面部图像特征的计算机自动定位方法。方法针对猕猴面部图像的特点建立其在HSI颜色空间中的肤色模型，然后根据各分量的阈值对图像进行分割，分割后的结果再经过数学形态学的腐蚀与膨胀处理得到猕猴面部较为完整的区域。结果基于面部器官位置的相对比例关系，实现了对猕猴眼睛、嘴巴等特征部位的自动定位，并且使用边缘提取算子获取了特征部位的细节轮廓。结论本文的算法经过了猕猴面部图像的测试，能准确的自动定位猕猴面部并找到眼部与嘴部区域，其结果可以为进一步的表情识别提供依据。Objective To propose an automatic computer method to localize facial features in macaque. Methods A color model is constructed in HSI color space based on the facial points of macaque. The threshold of the color model is applied to segment the macaque image. Then the comparative integrated facial image of the macaque can be obtained from the result by applying with erosion and dilation in mathematical morphology. Results Based on the comparative proportion relationship of the facial component＇s position, the macaques＇ facial components can be automatically located, which includes eyes, mouths and so on, it is also caught the detailed contour features of facial component with edge extraction methods. Conclusion The method demonstrated in this work is robust with quite high accuracy which can detect the facial components in macaques＇ facial image, and its result can support the further facial identification.

关 键 词：颜色空间 阈值分割 边缘提取 数学形态学 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]