INCORPORATING SPATIAL CONTEXT INTO STATISTICAL CLASSIFICATION OF MULTIDIMENSIONAL IMAGE DATA

JAMES CHARLES TILTON, Purdue University

Abstract

Compound decision theory is employed to develop a general statistical model for classifying image data using spatial context. The classification algorithm developed from this model exploits the tendency of certain ground-cover classes to occur more frequently in some spatial contexts than in others. A key input to this contextual classifier is a quantitative characterization of this tendency: the context function. Several methods for estimating the context function are explored, and two complimentary methods are recommended. The contextual classifier is shown to produce substantial improvements in classification accuracy compared to the accuracy produced by a non-contextual uniform-priors maximum likelihood classifier when these methods of estimating the context function are used. This improvement in classification accuracy is paid for by a substantial increase in computational requirements. An approximate algorithm, which cuts computational requirements by over one-half, is presented. Further reduction in computational requirements may be possible with a suggested hybrid algorithm. The search for an optimal implementation is furthered by an exploration of the relative merits of using spectral classes or information classes for classification and/or context function estimation. Finally, an unsuccessful attempt to devise a context measure for use in conjunction with context function estimation is described. Recommendations for further research are included in the concluding chapter.

Degree

Ph.D.

Subject Area

Electrical engineering

COinS