Expression Recognition Based on Multi-scale Block Local Gabor Binary Patterns with Dichotomy-Dependent Weights

Authors:
Zheng Zhang;Zheng Zhao;Tiantian Yuan
Affiliations:
College of Computer Science and Technology, Tianjin University, Tianjin, China 300072 and Tianjin University of Technology, Tianjin, China 300191;College of Computer Science and Technology, Tianjin University, Tianjin, China 300072;Tianjin University of Technology, Tianjin, China 300191
Venue:
ISNN 2009 Proceedings of the 6th International Symposium on Neural Networks: Advances in Neural Networks - Part II
Year:
2009

Citing 6
Cited 0

Multiresolution Gray-Scale and Rotation Invariant Texture Classification with Local Binary Patterns

IEEE Transactions on Pattern Analysis and Machine Intelligence
Coding Facial Expressions with Gabor Wavelets

FG '98 Proceedings of the 3rd. International Conference on Face & Gesture Recognition
Facial Expression Recognition Based on Gabor Wavelet Transformation and Elastic Templates Matching

ICIG '04 Proceedings of the Third International Conference on Image and Graphics
Facial Expression Recognition Based on Fusion of Multiple Gabor Features

ICPR '06 Proceedings of the 18th International Conference on Pattern Recognition - Volume 03
Automatic facial expression recognition using facial animation parameters and multistream HMMs

IEEE Transactions on Information Forensics and Security
Learning multi-scale block local binary patterns for face recognition

ICB'07 Proceedings of the 2007 international conference on Advances in Biometrics

Quantified Score

Hi-index	0.00

Visualization

Abstract

In order to accomplish subject-independent facial expression recognition, a weighted Multi-scale Block Local Gabor Binary Patterns (MB-LGBP) based facial expression recognition approach is presented in this paper. Gabor filters have been proved to be effective for expression recognition because of its superior capability of multi-scale representation, while MB-LBP is a powerful descriptor for encoding local-holistic textures. We combine the idea of Multi-scale Gabor representation with the concept of MB-LBP encoding to achieve both locally and globally informative MB-LGBP features. In recognition, we introduce dichotomy-dependent weights for SVM classification and compare its performance with the traditional weighted Chi square distance based paradigm. The promising result proves the superiority of the MB-LGBP composite features to some other popular features in expression recognition.