Non-classical Receptive Field Models with Inhibitory Subunits for High-speed Railway Image Enhancement

This paper presents two models for high-speed railway image enhancement. The basic model is inspired by some neuroelectrophysiological experiments on retinal ganglion cell's non-classical receptive field that reveals dis-inhibitory properties of inhibitory subunits. In view of the fact that there are different Gaussian channels in the model of receptive field, we improve the basic model to change the inhibitory effects by means of the potential function designed to show variations of inhibitory intensity adaptively. In order to validate the models, lots of highspeed railway image enhancement experiments are presented to show preferable results obtained by our models.