A Model of Saliency-Based Visual Attention for Rapid Scene Analysis
IEEE Transactions on Pattern Analysis and Machine Intelligence
Distinctive Image Features from Scale-Invariant Keypoints
International Journal of Computer Vision
Visual attention detection in video sequences using spatiotemporal cues
MULTIMEDIA '06 Proceedings of the 14th annual ACM international conference on Multimedia
ICNC '08 Proceedings of the 2008 Fourth International Conference on Natural Computation - Volume 06
A probabilistic model of overt visual attention for cognitive robots
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
Hi-index | 0.01 |
In this paper, we present a new spatiotemporal visual attention system. Typical feature integration model is expanded to incorporate motion in our suggested system, and is able to respond to motion stimulus by employing motion fields map as one of temporal features. Proposed system is based on bottom-up approach of human visual attention, but the main difference lies in its temporal feature extraction method, and integration method of multiple spatial and temporal features. Spatial features are integrated into spatial saliency map by weighted combination method. Temporal feature is extracted by SIFT and is analyzed and reorganized into temporal saliency map. Finally, dynamic fusion technique applied to make one spatiotemporal saliency map. To evaluate the performance of the system, we tested with various kinds of real video sequences. We also compared our system with several previous systems to validate the performance of the system.