基于曲率信息的鼻子区域精确定位  

Curvature-based morphological operators for automatic 3D nasal region segmentation

作　　者：郭振铎[1] 杨瑞敏[1] 刘洲峰[1] 高广帅 GUO Zhen-duo;YANG Rui-min;LIU Zhou-feng;GAO Guang-shuai(School of Electronic Inform ation, Zhongyuan U niversity of T echnology, Zhengzhou 450007,China)

机构地区：[1]中原工学院电子信息学院

出　　处：《计算机工程与设计》2017年第2期419-423,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(61202499;61379113);河南省基础与前沿技术研究计划基金项目(142300410042);郑州市科技领军人才基金项目(131PLJRC643)

摘　　要：提出一种三维人脸鼻子区域自动定位方法。采用一种基于近似平均曲率形态学分割算子来自动分割鼻子区域轮廓,该算子对种子的初始化位置具有很好的鲁棒性,可以对鼻子区域实现快速精确的分割。在国际通用的BU-3DFE人脸表情数据库进行测试,该方法对鼻子区域的分割达到了98.5%的准确率,验证了其有效性。A method for automatic localizing3D human face nose was proposed based on the approximate mean curvature and morphological operators.The proposed approach was robust to the initial position of the seed point,and achieved fast and accu-rate segmentation of the nose region.The experiments were conducted on the BU-3DFE dataset,and a final98.5%segmentation accuracy result was got.Experimental results show that the proposed approach can extract the nasal region accurately on the3D face.

关 键 词：曲率计算 曲率形态学算子 3D鼻子分割 BU-3DFE数据库 人脸识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]