ARTIFICIAL NEURAL-NETWORK RESPONSE TO MIXED PIXELS IN COARSE-RESOLUTION SATELLITE DATA

A feedforward neural network model based on the multilayer perceptron structure and trained using the backpropagation algorithm responds to subpixel class composition in both simulated and real data. Maps of th e network response surfaces for simulated data illustrate that the set of network outputs successfully characterizes the level of class domi nance and the subpixel composition for controlled data that contain a range of class mixtures. For a Sierra Nevada test site, the correspond ence between 250 m reference data and a network class map produced usi ng 250 m degraded TM data depends on the degree of subpixel class mixi ng as determined from coregistered 30 m reference data. For most misla beled pixels, classification error results from confusion between the first and second largest subpixel components, and the first and second largest network outputs. Overall map accuracy increases from 62% to 7 9% when mislabeled pixels are reclassified using the second largest ne twork output. Accuracy increases to 84% if, for mislabeled pixels, the second largest subpixel class is used as a reference. Maps of the net work response surfaces for a controlled subset of the Plumas data comp lement the findings of the simulated data and show that the network re sponds in a systematic way to changing proportions of subpixel compone nts. Based on our results we suggest that interpretation of the comple te set of network outputs can provide information on the relative prop ortions of subpixel classes. We outline a threshold-based heuristic th at would allow the labeling of pure classes, mixed classes, and primar y and secondary class types based on the relative magnitudes of the tw o largest network signals. (C) Elsevier Science Inc., 1996.