Extending color vision methods to bands beyond the visible
Machine Vision and Applications
A Hierarchical Approach to Robust Background Subtraction using Color and Gradient Information
MOTION '02 Proceedings of the Workshop on Motion and Video Computing
Fusion-Based Background-Subtraction using Contour Saliency
CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Workshops - Volume 03
Detecting moving objects, ghosts, and shadows in video streams
IEEE Transactions on Pattern Analysis and Machine Intelligence
Algorithms for a real-time HDR video system
Pattern Recognition Letters
| Hi-index | 0.00 |
In this paper, we present a multi-spectral video surveillance system. Improved background modeling and appearance-based object tracking are proposed with both signal-level and decision-level multi-spectral information fusion. In addition to modeling observations in each spectral channel by a typical pixel-level mixture-of-Gaussian-based model, we also model high level factors such as confidence of each modality, motion, object area, and lighting with a hierarchical probabilistic model feedback. This hierarchical model can enhance the performance of different challenging environment conditions, such as global illumination changes, and random parameter failures of background subtraction. Moreover, real-world vision problems include occlusion and merge/split are managed by our non-parametric tracking methodology and appearance-distance histogram. Our experiment in object tracking shows that under normal conditions, our system extends the capability of single spectral sensor, and under severe environment conditions, the overall system performance outperforms traditional direct fusion techniques in tracking reliability. This promising performance also encourages us to further extend our techniques to general multi-spectral and multi-modal surveillance.