Skip to main content
Figure 17 | BMC Neuroscience

Figure 17

From: A neural computational model for bottom-up attention with invariant and overcomplete representation

Figure 17

Comparison of different networks on illusory contour. (a) source image; (b) pooling result of fully connected network, which is almost a copy of input image since all features are reserved; (c) inhibition term of fully connected network which is almost uniform in the whole space; (d) saliency map of fully connected network; (e) two examples of pooling results of randomly connected network; (f) inhibition terms of randomly connected network corresponding to (e); (g) saliency map of randomly connected network; (h) four examples of pooling results of our model which describes features in specific orientations ; (i) inhibition terms of our model corresponding to (h); (j) saliency map of our model.

Back to article page