A noise deduction method for image data by using Centroid Neural Network (CNN) is proposed in this paper. The spatial constraints are applied to the Centroid Neural Network (CNN) algorithm to reduce noise in image data. The proposed algorithm incorporates a novel approach of using the weights of attributes to evaluate the roles of the latter. Experiments and results on a set of synthetic data show that the proposed algorithm provides a superior performance over other algorithms such as Fuzzy C-Means (FCM) and Fuzzy C-Means with spatial constraints (FCM-S).
展开▼