Application Research Of PCNN In Image Denoising And Face Recognition

With the development of social civilization, people use the computer not only hope to mathematical computation and information storage in the place of the human brain, but also more concerning the computer whether can mimic the brain for managing non-structure information, decision-making and intelligent controlling. The emergence of artificial neural network (ANN) to provide a great help for us to solve these problems, and become a kind of indispensable tools to study the field of artificial intelligence.The pulse coupled neural network (PCNN) as a representative of the third generation of the artificial neural network. It is better simulation the information processing mechanisms of the mammalian visual cortical neurons, and compared with the traditional neural network it has unique characteristics, such as double channel modulation characteristics, capture characteristics, similarity group neuronal pulse characteristics and dynamic threshold characteristics. The oscillation time sequences (OTS) of PCNN is unique and in a very small error conditions the translation, distortion, rotation and zoom is invariant, so the OTS can be used in feature description. Based on the many excellent characteristics of PCNN, now it has been widely used in image processing field.Firstly, this paper introduces the development of the ANN and the research status of PCNN, and describes the basic theory of PCNN in detail, then puts forward the simplified PCNN model, and briefly introduces the PCNN application in image processing. Secondly, it analyzes the insufficient of traditional filtering and the present situation of the PCNN filter to put forward an effective noise filtering method. This method use the characteristics of PCNN’s similar group of neurons released simultaneously pulse to locate the specific location of noise points, after the noise point located, using the filter composed of the morphological open and close operation to eliminate the noise points. The experiments show that this method is superior to the traditional filter in detail feature and edge information protection, and it can obtain good effect on high intensity noise pollution. Finally, several image preprocessing technology were introduced and described by experiments; and analysis the face recognition theory feasibility of PCNN (the oscillation time sequences of PCNN has the ability of characteristic description), then through the computer simulation experiment compared to PCA(principal component analysis), LDA(linear discriminate analysis), ICA (independent component analysis). it shows the practical feasibility that use PCNN to realize face recognition. For the image polluted by noise, proposed a method of face recognition with noise. In this method, firstly, using the proposed image de-noising method for preprocessing, and then using oscillation time sequences to feature extraction, finally using SVM for classification. The scientific nature and the validity of the method are proved by the MATLAB simulation experiment.