A bio-inspired model for color image segmentation

To segment an object from its background image for advanced vision processing, this article presents a novel bio-inspired framework for image segmentation in complex nature scenes, which is a hierarchical system that mimics the organization of layered early visual area in primate visual cortex. The proposed methodology consists of two typical stages: the first stage is a parallel modular structure including three segmenting operators based on color feature, form feature and texture feature, each of which solves the segmentation problem independently for the same input. They implement the similar computing as the parvocellular (P-cell), the magnocellular (M-cell) and koniocellular (K-cell) pathway in lateral geniculate nucleus (LGN) from the retina to the primary visual cortex. Then, a fusion operation, multiple feature fusion segmentation (MFFS), integrates these three feature segmentations together through the backpropagation neuron network (BPNN) in the last stage, which simulates the operation of area following the LGN in primary visual cortex. The proposed approach is applied to several segmentation experiments of many single objects in clustering conditions, the result shows that the approach is capable of competing with state-of-the-art systems.