VIU LogoVision & Image Understanding

Estimation of human-observer templates or perceptual filters (classification images)

Classification image in the spatial domain
(temporal sequence corresponds to classification images estimated based on increasing number of observer trials).

Classification image in fourier domain
(temporal sequence corresponds to classification images estimated based on increasing number of observer trials).

A brief introduction to classification images (published) 

Al Ahumada pioneered a technique known as classification images which allows investigators to infer how organisms use visual/auditory stimuli to make decisions. The technique uses external noise and the observers' decisions to estimate the underlying weights applied by the observer to the stimuli (first order classification image). Our work in this area has concentrated in developing methods and statistical techniques to estimate classification images for a variety of background noise statistics and apply the method to study interesting questions in cognition and perception.

On the methods side, we have extended the classification image technique to 2 alternative forced choice tasks (2 AFC) and to other than white noise backgrounds (filtered noise and real backgrounds).Two methods have been developed to estimate templates: 1) Unbiased estimator;  2) Maximum aposteriori probability.

We have used the technique to study a number of fundamental questions in perception and cognition: how the human perceptual filters change across a number of simple visual tasks (detection, contrast and size discrimination) in synthetic noise and real medical image backgrounds, how human observers use knowledge about the background statistics to adjust their strategy (templates or perceptual filters) in detection tasks, to study the underlying neural mechanisms mediating perception and eye movements, the integration of information driving saccades during search and visual attention in normal obsevers and hemineglect patients.


Journal of Vision special issue on classification images

Eckstein MP, Beutter BR, Pham BT, Shimozaki SS, Stone LS. Similar neural representations of the target for saccades and perception during search. J Neuroscience, 27, 1266-70, (2007)
For a comment on above article see: A Single Route to Action? The Common Representation of Perceptual and Saccade Targets by Thérèse Collins J. Neurosci. 2007 27: 3935-3936

Ludwig, C.H.J., Eckstein M.P., Brent R. Beutter, B.R., Limited flexibility in the filter underlying saccadic targeting, Vision Research, 47, 280-8, (2007)

Castella, C.,  Abbey, C.K.,  Eckstein, M.P., Verdun, F.R., Kinkel, K., Bochud, F.O., Human linear template with mammographic backgrounds estimated with genetic algorithm, Journal of the Optical Society of America A, 24, 1-12 (2007)

Shimozaki SS, Chen K, Abbey CK, Eckstein MP, The temporal dynamics of selective attention of the visual periphery as measured by classification images, Journal of Vision, 7(12):10, 1-20, (2007)

Shimozaki, S. S., Kingstone, A., Olk, B., Stowe, R., & Eckstein, M. P., Classification images of two right hemisphere patients: A window into the attentional mechanisms of spatial neglect, Brain Research, 1080, 26-52, (2006)

Eckstein, M.P., Pham B.T., Shimozaki, S.S., The footprints of visual attention during search with 100% valid and 100% invalid cues, Vision Research, 40, 1193-207 (2004)

Caspi, A, Beutter, B.R., Eckstein, M.P., The time course of visual information accrual guiding eye movement decisions, Proceedings of the National Academy of Sciencies,101: 13086-13090 (2004)

Eckstein, M. P., Shimozaki, S. S., & Abbey, C. K. The footprints of visual attention in the Posner cueing paradigm revealed by classification images. Journal of Vision, 2(1), 25-45,, DOI 10.1167/2.1.3, (2002)

Abbey, C. K. & Eckstein, M. P.Classification image analysis: Estimation and statistical inference for two-alternative forced-choice experiments. Journal of Vision, 2(1), 66-78,, DOI 10.1167/2.1.5 (2002)

Abbey, C.K., Eckstein, M.P., Optimal shifted estimates of human-observer templates in two-alternative forced choice experiments, IEEE Transactions on Medical Imaging, 21, 429-440, (2002)

C.K. Abbey and M.P. Eckstein, "Theory for estimating human-observer templates in two-alternative forced-choice experiments" To be published in: Proc. 17th Int. Conf. onInformation Processing in Medical Imaging (M.F. Insana and R. Leahy, Eds.), Springer-Verlag, Berlin, 2001.

C.K. Abbey and M.P. Eckstein, "Maximum-likelihood and maximum a-posteriori estimates of human-observer templates," Proc. SPIE (E.A. Krupinski and D.P. Chakraborty,Ed.s) 4325, 2001

C.K. Abbey and M.P. Eckstein, "Estimates of human-observer templates for simple detection tasks in correlated noise," Proc. SPIE (E.A. Krupinski, Ed.) 3981:70-77, 2000.

C.K. Abbey, M.P. Eckstein, and F.O. Bochud, "Estimation of human-observer templates for 2 alternative forced choice tasks",  Proc. SPIE (E.A. Krupinski, Ed.) 3663:284-295, 1999.