Robust Image Segmentation Using Pulse-coupled neural network with De-noising by Kernel PCA

An image segmentation algorithm based on pulse-coupled neural network with de-noising by kernel principal component analysis (PCA) is presented in this paper. PCA enables us to extract nonlinear features and therefore performs as a powerful preprocessing step for classification. But the extracted feature components are sensitive to outliers contained in data. This is a characteristic common to all PCA-based techniques. In this paper, the kernel PCA is proposed, which is able to remove outliers in data vectors and achieve dimension reduction. After de-noising, a new image segmentation approach based on pulse coupled neural network (PCNN) is presented. PCNN dynamically evaluates similarity between any two samples owing to the outstanding centralization characteristic based on the vicinity in space and the comparability of brightness. It has higher accuracy and faster performance than those classical clustering algorithms. The pulse coupled neural network clustering algorithm with de-noising by kernel PCA enhances robustness of the original clustering algorithms to noise. Experimental results with artificial and real-world images have shown the effectiveness of the proposed algorithm