|
|||||
|
|
| 基于类灰度图的类Haar特征构建及其应用 | |
| 引用本文： | 陈锻生,陈齐松,刘政凯.基于类灰度图的类Haar特征构建及其应用[J].郑州大学学报(理学版),2007,39(1):33-39. |
| 作者姓名： | 陈锻生 陈齐松 刘政凯 |
| 作者单位： | 1. 华侨大学计算机科学系,福建,泉州,362021;中国科学技术大学电子工程与信息科学系,合肥,230027 2. 华侨大学计算机科学系,福建,泉州,362021 3. 中国科学技术大学电子工程与信息科学系,合肥,230027 |
| 摘 要： | 有效的特征提取是正确模式分类的基础.在Viola-Jones快速目标检测算法的基础上,侧重研究了类Haar特征原型的本质与提取,提出了类灰度图的概念,并以快速人脸检测为例,从类灰度图提取广义类Haar特征,从本质上拓展了类Haar特征的类型.基于类灰度图提取的广义类Haar特征在基于Adaboost的人脸检测系统性能超过Viola-Jones系统;作为弱分类器的滤波器构建是影响基于Adaboost分类算法性能的重要因素. |
| 关 键 词： | 类灰度图 类Haar特征 特征提取 人脸检测 |
| 文章编号： | 1671-6841(2007)01-0033-07 |
| 修稿时间： | 05 15 2006 12:00AM |
Construction and Application of Haar-like Features Based on Gray-like Images |
|
| CHEN Duan-sheng,CHEN Qi-song,LIU Zheng-kai.Construction and Application of Haar-like Features Based on Gray-like Images[J].Journal of Zhengzhou University:Natural Science Edition,2007,39(1):33-39. | |
| Authors: | CHEN Duan-sheng CHEN Qi-song LIU Zheng-kai |
| Affiliation: | 1. Department of Computer Science, Huaqiao University, Quanzhou 362021, China ; 2. Department of Electronic Engineering and Information Science, University of Science and Technology of China, Hef ei 230027, China |
| Abstract: | In pattern classification,correct classification is based on effective feature extraction.Based on the Viola-Jones fast object detection algorithm, the nature and the construction of Haar-like features are researched.A new concept of gray-like images is put forward.As an example of the gray-like images application,the generalized Haar-like features can be constructed for fast face detection,and the Haar-like feature types are expanded essentially.The Adaboost based on face detector using the generalized Haar-like features is superior to the Viola-Jones' system.The construction of filters for weak classifiers is one of the most important factors for Adaboost based on classification system. |
| Keywords: | gray-like image Haar-like feature feature extraction face detection |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ | |
