Facial occlusion reconstruction: recovering both the global structure and the local detailed texture components

Authors:
Ching-Ting Tu;Jenn-Jier James Lien
Affiliations:
Robotics Laboratory, Department of Computer Science and Information Engineering, National Cheng Kung University, Tainan, Taiwan, R.O.C.;Robotics Laboratory, Department of Computer Science and Information Engineering, National Cheng Kung University, Tainan, Taiwan, R.O.C.
Venue:
PSIVT'07 Proceedings of the 2nd Pacific Rim conference on Advances in image and video technology
Year:
2007

Citing 9
Cited 1

Matrix computations (3rd ed.)

Matrix computations (3rd ed.)
Linear Object Classes and Image Synthesis From a Single Example Image

IEEE Transactions on Pattern Analysis and Machine Intelligence
Multidimensional Morphable Models: A Framework for Representing and Matching Object Classes

International Journal of Computer Vision
Reconstruction of Partially Damaged Face Images Based on a Morphable Face Model

IEEE Transactions on Pattern Analysis and Machine Intelligence
Learning Low-Level Vision

ICCV '99 Proceedings of the International Conference on Computer Vision-Volume 2 - Volume 2
Face Identification across Different Poses and Illuminations with a 3D Morphable Model

FGR '02 Proceedings of the Fifth IEEE International Conference on Automatic Face and Gesture Recognition
Person identification from heavily occluded face images

Proceedings of the 2004 ACM symposium on Applied computing
Automatic Eyeglasses Removal from Face Images

IEEE Transactions on Pattern Analysis and Machine Intelligence
Glasses Removal from Facial Image Using Recursive Error Compensation

IEEE Transactions on Pattern Analysis and Machine Intelligence

Reconstruction of occluded facial images using asymmetrical Principal Component Analysis

Integrated Computer-Aided Engineering

Quantified Score

Hi-index	0.00

Visualization

Abstract

An automatic facial occlusion reconstruction system based upon a novel learning algorithm called the direct combined model (DCM) approach is presented. The system comprises two basic DCM modules, namely a shape reconstruction module and a texture reconstruction module. Each module models the occluded and non-occluded regions of the facial image in a single, combined eigenspace, thus preserving the correlations between the geometry of the facial features and the pixel grayvalues, respectively, in the two regions. As a result, when shape or texture information is available only for the nonoccluded region of the facial image, the optimal shape and texture of the occluded region can be reconstructed via a process of Bayesian inference within the respective eigenspaces. To enhance the quality of the reconstructed results, the shape reconstruction module is rendered robust to facial feature point labeling errors by suppressing the effects of biased noises. Furthermore, the texture reconstruction module recovers the texture of the occluded facial image by synthesizing the global texture image and the local detailed texture image. The experimental results demonstrate that compared to existing facial reconstruction systems, the reconstruction results obtained using the proposed DCM-based scheme are quantitatively closer to the ground truth.