| 基于特征点定位与肤色分割的胡子检测算法 |
| |
| 引用本文： | 蔡晓东,甘凯今,杨超,王丽娟,王迪. 基于特征点定位与肤色分割的胡子检测算法[J]. 电视技术, 2016, 40(3): 116-121. DOI: 10.16280/j.vide0e.2016.03.025 |
| |
| 作者姓名： | 蔡晓东 甘凯今 杨超 王丽娟 王迪 |
| |
| 作者单位： | 1. 桂林电子科技大学信息与通信学院,广西桂林,541004;2. 桂林电子科技大学机电工程学院,广西桂林,541004 |
| |
| 基金项目： | 国家科技支撑计划课题；广西科学研究与技术开发计划项目；桂林电子科技大学研究生教育创 新计划资助项目. |
| |
| 摘 要： | 针对具有一定程度偏转和模糊的人脸图像难以实现胡子检测的问题,提出基于人脸特征点定位与肤色分割的胡子检测算法.该算法的设计思路是首先使用主动形状模型(ASM)算法定位人脸特征点进而获取下巴区域,然后利用提出的自动选择聚类中心(ASCC)的肤色分割算法分离出下巴的非肤色区域,最后在下巴非肤色区域中使用胡子颜色判别法检测得到胡子.在LFW人脸库上的实验表明,该算法能够准确地检测出入脸的胡子,特别地,对有一定程度偏转和模糊的人脸图像,算法依然能获取良好的检测效果.
|
| 关 键 词： | 胡子检测 主动形状模型 人脸特征点 肤色分割 |
| 收稿时间： | 2015-07-09 |
| 修稿时间： | 2015-09-15 |
Beard Detection Based on Facial Feature Points Location and Skin Color Segmentation |
| |
| cai xiao dong,gan kai jin,Yang Chao,wang li juan and Wang Di. Beard Detection Based on Facial Feature Points Location and Skin Color Segmentation[J]. Ideo Engineering, 2016, 40(3): 116-121. DOI: 10.16280/j.vide0e.2016.03.025 |
| |
| Authors: | cai xiao dong gan kai jin Yang Chao wang li juan Wang Di |
| |
| Affiliation: | School of Information and Communication of Guilin University of Electronic Technology,Guangxi Guilin,School of Information and Communication of Guilin University of Electronic Technology,Guangxi Guilin,School of Information and Communication of Guilin University of Electronic Technology,Guangxi Guilin,School of Information and Communication of Guilin University of Electronic Technology,Guangxi Guilin,Electromechanical Engineering College of Guilin University of Electronic Technology,Guangxi Guilin |
| |
| Abstract: | In order to solve the problems which beard/mustache detection may encounter when using profile or fuzzy face image, this paper propose a beard/mustache detection algorithm that combined with facial feature points location and color segmentation. First, we employ the Active Shape Model (ASM) algorithm to mark facial feature points and then get the chin area. Secondly, we use the proposed skin segmentation algorithm of Auto Selected Clustering Center (ASCC) to separate chin area into skin color and non-skin color region. Finally, beard/mustache can be detected in non-skin color region of chin by using the beard color discriminance. The experimental results on LFW database have shown that the propose beard/mustache detection algorithm is practical and effective. Especially, in the case of profile or fuzzy face image, the approach can still obtain a good result. |
| |
| Keywords: | beard detection Active Shape Model facial feature points skin color segmentation |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《电视技术》浏览原始摘要信息 |
|
点击此处可从《电视技术》下载全文 |
|
