Cortical Surround Interactions and Perceptual Salience via Natural Scene Statistics
Figure 3
Center-surround configurations corresponding to the different mixture components of the model.
We used a bank of linear filters, depicted as colored bars in the top row, comprising 4 orientations at 1 center position and 8 surround positions. We grouped surround RFs according to their orientation, each labeled with a different color. A surround group can be either co-assigned with the center group (i.e., the model assumes dependence between center and surround groups, and includes them both in the normalization pool for the center, as in Fig. 2-top), or not co-assigned (i.e. the model assumes independence between center and surround groups, and does not include the latter in the normalization pool, as in Fig. 2-bottom). The leftmost column depicts the configuration (denoted by ) in which none of the surrounds is co-assigned with the center;
best describes image patches such as those identified by the circles in the bottom row (red and orange circles denote center and surround respectively, as in Fig. 1). The second column depicts the configuration (denoted by
) in which the vertical surround is co-assigned with the center; black bars (top) identify the co-assigned groups, circles (bottom) the image patches best described by
. The same conventions are used in the remaining columns.