An improved PCNN model and a new removing algorithm of salt and pepper noise

Abstract

An improved PCNN model-PCNN with Positive and Negative Firing, PCNNPNF-is proposed, and also put forward a de-noising algorithm based on the time matrix of PCNNPNF. The biggest improvement is that the neuron's output of improved PCNN has three states: positive firing, negative firing and no firing, while PCNN only has two states: firing and no firing. Experimental results show that the de-noising algorithm based on PCNNPNF can quickly find the two kinds of pulse noises, remove these noises, and reserve more information than PCNN.