This work deals with multisensor data fusion to obtain landcover classification. The role of feature-level fusion using the Dempster-Shafer rule and that of data-level fusion in the MRF context is studied in this paper to obtain an optimally segmented image. Subsequently, segments are validated and classification accuracy for the test data is evaluated. Two examples of data fusion of optical images and a synthetic aperture radar image are presented, each set having been acquired on different dates. Classification accuracies of the technique proposed are compared with those of some recent techniques in literature for the same image data.

An unsupervised segmentation approach to classification of multispectral image is suggested here in Markov random field (MRF) frame work. This work generalizes the work of Sarkar et al. (2000) on gray value images for multispectral images and is extended for landuse classification. The essence of this approach is based on capturing intrinsic characters of tonal and textural regions of any multispectral image. The approach takes an initially oversegmented image and the original. multispectral image as the input and defines a MRF over region adjacency graph (RAG) of the initially segmented regions. Energy function minimization associated with the MRF is carried out by applying a multivariate statistical test. A cluster validation scheme is outlined after obtaining optimal segmentation. Quantitative evaluation of classification accuracy of test data for three illustrations are shown and compared with conventional maximum likelihood procedure. Comparison of the proposed methodology with a recent work of texture segmentation in the literature has also been provided. The findings of the proposed method are found to be encouraging.

A simple technique has been suggested to obtain optimal segmentation based on tonal and textural characteristics of an image using the Markov random field (MRF) model. The technique takes an initially over segmented image as well as the original image as its inputs and defines an MRF over the region adjacency graph (RAG) of the initially segmented regions. A tonal-region based segmentation technique due to Kartikeyan and Sarkar (1989) has been used for initial segmentation. The energy function has been defined over the first order cliques of the MRF. The essence of this approach is primarily based on quantitative values of the second order statistics, on region characteristics and consequently deciding upon the action of merging neighboring regions using the F-statistic. The effectiveness of our approach is demonstrated with wide variety of real life examples viz., indoor, outdoor and satellite and a comparison of its output with that of a previous work in the literature has been provided.

This paper addresses the problem of identification of appropriate autoregressive (AR) components to describe textural regions of digital images by a general class of two-dimensional (2-D) AR models. In analogy with univariate time series, the proposed technique first selects a neighborhood set of 2-D lag variables corresponding to the significant multiple partial auto-correlation coefficients. A matrix is then suitably formed from these 2-D lag variables. Using singular value decomposition (SVD) and orthonormal with column pivoting factorization (QRcp) techniques, the prime information of this matrix corresponding to different pseudoranks is obtained. Schwarz's (1978) information criterion (SIG) is then used to obtain the optimum set of 2-D lag variables, which are the appropriate autoregressive components of the model for a given textural image. A four-class texture classification scheme is illustrated with such models and a comparison of the technique with the work of Chellappa and Chatterjee (1985) is provided.