G. Azzopardi and N. Petkov, “A CORF computational model of a simple cell with application to contour detection”, AVA/BMVA Meeting on Biological and Computer Vision, Microsoft Research Center, Cambridge, UK, May 2012
We propose a computational model of a simple cell in primary visual cortex that relies on input from model LGN cells with center-surround receptive fields (RFs) and we refer to it as a Combination of RFs (CORF) model. We compare the proposed CORF model with the 2D Gabor function (GF) model that has gained particular popularity as a computational model of a simple cell. The GF model bypasses the LGN and its effectiveness in contour detection tasks has never been compared with the effectiveness of alternative models. The RF map of the proposed CORF model, determined with simulated reverse correlation, can be divided in elongated excitatory and inhibitory regions typical of simple cells. Besides orientation selectivity, the CORF model achieves modulated response to shifted gratings, and exhibits cross orientation suppression, contrast invariant orientation tuning and response saturation. These properties are observed in real simple cells but are not possessed by the GF model. The proposed CORF model also outperforms the GF model, with high statistical confidence, in a contour detection task on two public data sets of images of natural scenes with associated contour ground truths: RuG data set (t(39) = 4.39, p < 10-4) and Berkeley data set (t(299) = 3.88, p < 10-4). The proposed CORF model is more realistic than the GF model as it shares more properties with real simple cells and it is more effective in contour detection, which is assumed to be the primary biological role of simple cells.