Adaptive color transformation for person re-identification in camera networks

Authors:
Clemens Siebler;Keni Bernardin;Rainer Stiefelhagen
Affiliations:
Karlsruhe Institute of Technology, Karlsruhe, Germany;Karlsruhe Institute of Technology, Karlsruhe, Germany;Karlsruhe Institute of Technology, Karlsruhe, Germany
Venue:
Proceedings of the Fourth ACM/IEEE International Conference on Distributed Smart Cameras
Year:
2010

Citing 6
Cited 0

Improved Adaptive Gaussian Mixture Model for Background Subtraction

ICPR '04 Proceedings of the Pattern Recognition, 17th International Conference on (ICPR'04) Volume 2 - Volume 02
Appearance Modeling for Tracking in Multiple Non-Overlapping Cameras

CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 2 - Volume 02
Tracking people across disjoint camera views by an illumination-tolerant appearance representation

Machine Vision and Applications
Modeling inter-camera space-time and appearance relationships for tracking across non-overlapping views

Computer Vision and Image Understanding
Viewpoint Invariant Pedestrian Recognition with an Ensemble of Localized Features

ECCV '08 Proceedings of the 10th European Conference on Computer Vision: Part I
Tracking objects across cameras by incrementally learning inter-camera colour calibration and patterns of activity

ECCV'06 Proceedings of the 9th European conference on Computer Vision - Volume Part II

Quantified Score

Hi-index	0.00

Visualization

Abstract

Cross-camera matching is often done using color features, which can compensate poses and viewpoint changes fairly well. On the downside, color features are very sensitive to illumination changes. In multi-camera systems, cameras are often installed at different physical sites and therefore illumination conditions are likely to differ, which highly influences the accuracy of the matching process. To overcome this mismatch, a Brightness Transfer Function can be trained to establish a mapping of brightness values between the different views. However, as soon as the lighting conditions do not match the training condition any longer, mapping accuracy tends to decrease again. In this work, an unsupervised approach for mitigating the effects of intra-camera illumination change is proposed. After applying such a procedure, the changing illumination over time has less impact on the fixed training stage and therefore inter-camera recognition performance improves.