Simulated light sensitive model for handwritten digit recognition

A new technique for handwritten digit recognition based on a simulated light sensitive model combined with the classical back-propagation network is proposed. The primary objective is to overcome the space and time complexities of the problem. The ultimate goal is to implement the recognition technique on a VLSI chip. The authors discuss the recognition concept and introduce the light receptor model. The excitation and inhibition rules are considered. The experiments confirmed that this technique substantially reduces the time and space complexities of the network. It is better than the existing techniques