计算机工程 ›› 2011, Vol. 37 ›› Issue (18): 198-200.doi: 10.3969/j.issn.1000-3428.2011.18.066

• 人工智能及识别技术 • 上一篇    下一篇

基于生物启发特征的真实环境笑脸分类方法

陈 俊   

  1. (华南理工大学电子与信息学院,广州 510640)
  • 收稿日期:2011-03-14 出版日期:2011-09-20 发布日期:2011-09-20
  • 作者简介:陈 俊(1979-),男,讲师、博士研究生,主研方向:智能信息处理

Classification Method of Smiling Face in Real Environment Based on Biologically Inspired Feature

CHEN Jun   

  1. (School of Electronic and Information, South China University of Technology, Guangzhou 510640, China)
  • Received:2011-03-14 Online:2011-09-20 Published:2011-09-20

摘要: 为解决生物启发模型(BIM)存在的3个问题,即高计算复杂度、有争议的视觉皮层关系建模,以及类前向反馈机制带来的盲目特征选择,提出一种基于生物启发特征(BIF)的真实环境笑脸分类方法。构建基于BIF的笑脸分类系统,提取人脸表情图像嘴部区域的金字塔梯度方向直方图特征,使用局部保持投影进行BIM特征降维,采用Adaboost算法进行BIM特征选择。实验结果验证,该系统的最佳识别率达96.5%。

关键词: 笑脸表情分类, 生物启发特征, 金字塔梯度方向直方图特征, 局部保持投影, 支持向量机

Abstract: To deal with the following three issues: the high computational complexity, the controversial modeling on the visual cortex relationship, and the blind feature selection due to the characteristics of similar feed-forward framework that exists in Biologically Inspired Model(BIM), this paper proposes to build a smiling face classification system based on BIM. It extracts Pyramid Histogram of Oriented Gradients(PHOG) features form the mouth region of the facial expression image, reduces the BIM feature vector dimensions using Locality Preserving Projection(LPP), and adopts Adaboost algorithm for BIM feature selection. Experimental results demonstrate that the best recognition rate is up to 96.5%.

Key words: smiling face expression classification, Biologically Inspired Feature(BIF), Pyramid Histogram of Oriented Gradients(PHOG) feature, Locality Preserving Projection(LPP), Support Vector Machine(SVM)

中图分类号: