Cascade MR-ASM for locating facial feature points

Authors:
Sicong Zhang;Lifang Wu;Ying Wang
Affiliations:
School of Electronic Information and control Engineering, Beijing University of Technology, Beijing, China;School of Electronic Information and control Engineering, Beijing University of Technology, Beijing, China;School of Electronic Information and control Engineering, Beijing University of Technology, Beijing, China
Venue:
ICB'07 Proceedings of the 2007 international conference on Advances in Biometrics
Year:
2007

Citing 3
Cited 2

View-Based Active Appearance Models

FG '00 Proceedings of the Fourth IEEE International Conference on Automatic Face and Gesture Recognition 2000
Learning A Single Active Face Shape Model across Views

RATFG-RTS '99 Proceedings of the International Workshop on Recognition, Analysis, and Tracking of Faces and Gestures in Real-Time Systems
Manifold Based Analysis of Facial Expression

CVPRW '04 Proceedings of the 2004 Conference on Computer Vision and Pattern Recognition Workshop (CVPRW'04) Volume 5 - Volume 05

Coarse-to-fine statistical shape model by Bayesian inference

ACCV'07 Proceedings of the 8th Asian conference on Computer vision - Volume Part I
A fuzzy vault scheme for feature fusion

CCBR'11 Proceedings of the 6th Chinese conference on Biometric recognition

Quantified Score

Hi-index	0.00

Visualization

Abstract

Accurate and robust location of feature point is a difficult and challenging issue in face recognition. In this paper we propose a new approach of using a cascade of Multi-Resolution Active Shape Models (C-MR-ASM) to locate facial feature points. In our approach, more than one MR-ASMs are obtained from different subsets of training set automatically, and these MR-ASMs are integrated in a cascade to locate facial feature points. Experimental results show that our algorithm is more accurate than traditional MR-ASM. The contribution of this paper includes: 1, unlike traditional MR-ASM, the training set is divided into several subsets automatically based on the principle a trained model should describe all the samples in training set accurately. 2, we propose the new cascade framework, which integrates all the subset MR-ASM.