Spatio-temporal Saliency Detection:

Note:

1. I use Saliencytoolbox (STB) 2.1 of [1]. The parameters are set according to Dr. Dirk's suggestion.

2. I use Neuromorphic Vision Toolkit (NVT) as the realization of Itti's work [2].

3. SR FFT represents [3]'s method.

4. PQFT and PFT are my proposed methods [4].

5. All the experiments were run in Matlab 2007a on Linux platform.

I. Video Sequences (988 images captured at 15f/s): Download [3.49M]
Model PQFT PFT SR NVT STB	Average Time Cost (s) 0.0565 0.0106 0.0141 0.4313 3.5337
II. Natural Images (100 images in [1,2,3]):
Model PQFT PFT SR NVT STB	Average Time Cost (s) 0.0597 0.0099 0.0159 0.7440 4.7395
III. Natural Images with white-colored noise (the image from [2]):
IV. Psychological Patterns:
1. Color and Orientation Patterns.
2. Only Orientation Patterns.
3. Grouping search
4. Conjunction search
More demos will appear soon! If you have any question, please send a mail to chenlei.guo@gmail.com. You can find my past work at my personal website: http://homepage.fudan.edu.cn/~clguo.   Reference: [1] D. Walther and C. Koch (2006), Modeling attention to salient proto-objects. Neural Networks 19, 1395-1407. [2] L. Itti, C. Koch, and E. Niebur. (1998). A model of saliency-based visual attention for rapid scene analysis. IEEE Trans. on PAMI, 20(11), 1254-1259 [3] X. Hou and L. Zhang (2007). Saliency Detection: A Spectral Residual Approach. CVPR 2007. [4] C.L. Guo, Q. Ma and L.M. Zhang. (2008). Spatio-temporal Saliency Detection Using Phase Spectrum of Quaternion Fourier Transform, CVPR 2008.