Techniques for fine resolution MaxEnt image analysis

Abstract

The Maximum Entropy Method (MaxEnt) is a Bayesian technique for the reconstruction of images from imperfect or noisy data. The development of MaxEnt is discussed. MaxEnt has been applied in a wide variety of circumstances including X-ray astronomy (Gull and Daniell (1978)), astronomy (Loredo (1990)), spectral analysis (Lizamore (1995)), positron emission tomography (O'Sullivan (1995)) and virtually all types of image processing. MaxEnt is applied to the task of estimating a fishery density from trawl catch data. The data relates to the Hoki fishery off the West Coast of the South Island, New Zealand. A tomographic approach is developed. In this situation the source is the fish density and the trawl lines represent cross-sectional scan lines. The use of reliability (straightness) information, spatial correlations and acoustic survey data has been developed to produce superior images. The results are consistent with prior information about the dynamics of the fishery but provide higher accuracy than obtained previously.