Abstract

T. N. Mundhenk, L. Itti, CINNIC, a new computational algorithm for the modeling of early visual contour integration in humans, Neurocomputing, Vol. 52-54, pp. 599-604, Jun 2003. [2001 impact factor: 0.534] (Cited by 10)

Abstract: We have developed a computational model called CINNIC to simulate contour integration and visual salience in early visual processing in the human brain. Our model uses the standard butterfly pattern of connections between early orientation selective neurons, which are believed to mediate interactions along contour elements. However, we add multi scale analysis, adaptive neuron group suppression and fast plasticity of connection weights to increase the performance of our algorithm. We show quantitatively that the addition of these ideas helps in the detection of salient contours. We also submit that our algorithm is biologically plausible and falls in line with what is known about neuron connections and interactions in V1 and possibly V2.

Themes: Computational Modeling, Model of Bottom-Up Saliency-Based Visual Attention