Interval Type-2 Fuzzy C-Means Clustering with Spatial Information for Land-Cover Classification

Abstract The paper proposes a method to use spatial information to interval type-2 fuzzy c-Means clustering (IT2FCM) for problems of land cover classification from multi-spectral sattelite images. The spatial information between a pixel and its neighbors on individual band is used to calculate an interval of membership grades in IT2-FCM algorithm. The proposed algorithm, called IIT2-FCM, is implemented on Landsat7 images in comparison with previous algorithms like k-Means, FCM, IT2-FCM to demonstrate the advantage of the approach in handling uncertainty or noise. Keywords (separated by '-') Fuzzy clustering - Type-2 fuzzy sets - Land cover classification - Spatial information 

• Type: Journal
• Year: 2015
• Where: ACIIDS
• Authors: Dinh Sinh Mai, Long Thanh Ngo
• Keywords: ACIIDS 2015, Database