Super-resolution mapping of multiple-scale land cover features using a Hopfield neural network

Soft classification techniques have been developed to estimate the class composition of image pixels, but their output provides no indication of how these classes are distributed spatially within the pixel. Separate Hopfield neural network techniques for producing super-resolution maps from imagery of features larger and smaller than a pixel have been developed. However, the techniques have yet to be combined in order to produce super-resolution maps of multiple-scale land cover features. This paper presents the first results from combining the two approaches. The output from a soft classification and prior information of sub-pixel feature arrangement is used to constrain a Hopfield neural network formulated as an energy minimisation tool. The energy minimum represents a 'best guess' map of the spatial distribution of class components in each pixel. The technique was applied to simulated SPOT HRV imagery and the resultant maps provided an accurate and improved representation of the land covers studied.