A neural circuit model for nCRF's dynamic adjustment and its application on image representation

Abstract

According to Biology there is a large disinhibitory area outside the classical receptive field (CRF), which is called as non-classical receptive field (nCRF). Combining CRF with nCRF could increase the sparseness, reliability and precision of the neuronal responses. This paper is aimed at the realization of the neural circuit and the dynamic adjustment mechanism of the receptive field (RF) with respect to nCRF. On the basis of anatomical and electrophysiological evidence, we constructed a neural computational model, which can represent natural images faithfully, simply and rapidly. And the representation can significantly improve the subsequent operation efficiency such as segmentation or integration. This study is of particular significance in the development of efficient image processing algorithms based on neurobiological mechanisms. The RF mechanism of ganglion cell (GC) is the result of a long term of evolution and optimization of self-adaptability and high representation efficiency. So its performance evaluation in natural image processing is worthy of further study.