Satellite Image Classification Based Spatial-Spectral Fuzzy Clustering Algorithm

Abstract

Spectral clustering is a clustering method based on algebraic graph theory. The clustering effect by using spectral method depends heavily on the description of similarity between instances of the datasets. Althought, spectral clustering has gained considerable attentions in the recent past, but the raw spectral clustering is often based on Euclidean distance, but it is impossible to accurately reflect the complexity of the data. Despite having a well-defined mathematical framework, good performance and simplicity, it suffers from several drawbacks, such as it is unable to determine a reasonable cluster number, sensitive to initial condition and not robust to outliers. Owing to the limitations of the feature space in multispectral images and spectral overlap of the clusters, it is required to use some additional information such as the spatial context in image clustering. In this paper, we present a new approach named spatial-spectral fuzzy clustering (SSFC) which combines spectral clustering and fuzzy clustering with local information into a unified framework to solve these problems and also using fuzzy clustering algorithm to converge the global optimization, this method is simple in computation but quite effective when solving segmentation problems on satellite imagery. Making it to find the spatial distribution characteristics of complex data and can further make cluster more stable. Experimental results show that it can improve the clustering accuracy and avoid falling into local optimum.