A heuristic for the maximum value region problem in raster space

• From a single-attribute raster layer in which each cell is assigned a numerical value, a connected set of a specified number of cells that has the maximum (or minimum) total value is selected. This is a highly common decision problem in the context of raster-based geographic information systems (GIS) and seems general enough to deserve inclusion in the standard functionality of such systems. Yet it is a computationally difficult optimization problem, for which no efficient exact solution method has been found. This article presents a new dynamic programming-based heuristic method for the problem. Its performance is tested with randomly generated raster layers with various degrees of spatial autocorrelation. Results suggest that the proposed heuristic is a promising alternative to the existing integer programming-based exact method, as it can handle significantly larger raster data with fair accuracy.

Ämnesord

TEKNIK OCH TEKNOLOGIER  -- Naturresursteknik -- Geofysisk teknik (hsv//swe)

ENGINEERING AND TECHNOLOGY  -- Environmental Engineering -- Geophysical Engineering (hsv//eng)

Nyckelord

decision-support system

landscape ecology

field-object representation

raster data modeling

Publikations- och innehållstyp

ref (ämneskategori)

art (ämneskategori)