The Modified Pulse Coupled Neural Networks For The Application Of Mixed Noise In Images

Pulse coupled neural network (PCNN) is a new generation artificial neural network which has a better biological characteristics and is different from the traditional artificial neural network. The PCNN neuron has the phenomenon of impulse emission and the ability to capture similar neurons to pulse at the same time, has the variable threshold to make the dissimilarity neurons to pulse at the different time and the characteristics of space-time accumulation. With the similar biological characteristics of the visual cell, PCNN has a unique advantage in image processing.The dissertation did further research and brought the simplified PCNN and improved PCNN based on the traditional PCNN model. It was used in the filtering algorithm of impulse noise and mixed noise.The primary research works of the dissertation are as follows:First, the dissertation summarized the PCNN model and the principle and the simplified PCNN model and improved PCNN model.Second, the dissertation introduced the noise model and some filtering methods.Third, combining the simplified PCNN and median filter to remove the pulse noises can solve the blindness of the traditional noise filter. Using the characteristics of the pulse capturing and space-time accumulation to locate the noise and removing can obtain a good filtering result.Forth, using the improved PCNN model to remove the mixed noises can solve the flaw of the threshold output model and create a filtering algorithm with the weighted median output. It can sort the noises and remove it, and then has a good result.