J4

• 计算机科学 • 上一篇    下一篇

小波和神经网络在人脸光照校正中的应用

张忠波, 马驷良, 马捷   

  1. 吉林大学数学研究所, 长春 130012
  • 收稿日期:2004-06-23 修回日期:1900-01-01 出版日期:2005-03-26 发布日期:2005-03-26
  • 通讯作者: 马驷良

Application of Wavelet and Neural Networkin the Human Face Illumination Compensation

ZHANG Zhong-bo, MA Si-liang, MA Jie   

  1. Institute of Mathematics, Jilin University, Changchun 130012, China
  • Received:2004-06-23 Revised:1900-01-01 Online:2005-03-26 Published:2005-03-26
  • Contact: MA Si-liang

摘要: 提出一种基于小波和神经网络的人脸光照校正算法, 它 充分结合了小波的多分辨分析和BP神经网络的自适应学习和良好推广能力的性质, 对人脸进行光照校正. 理论和实验证明, 本方法能有效解决人脸检测和识别中的光照校正问题, 使不同光照条件下的人脸检测和人脸识别性能得到很大提高. 本方法还具有鲁棒性好、 适应面宽的特点.

关键词: 光照校正, 小波分析, 神经网络, 人脸识别, 人脸检测

Abstract: Based on wavelet and neural network the method of face illumination compensation is proposed. It sufficiently combines multiresolution analysis of wavelet with the self adaptation learning and ability of good spread of BP neural network, thus this method carries out the face illumination compensation. The theory and experiments prove that this method solves face detection and illumination compensation in the identifying process. Moreover, it has fine robust and adaptable characteristics in a wide range.

Key words: illumination compensation, wavelet analyze, neural net work, face recognition, face detection

中图分类号: 

  • TP301