Selective image enhancement using attribute morphology

Abstract

Numerous applications in image interpretation involve feature extraction. In many low-contrast images, this process is pre-empted by an image enhancement phase to improve the quality of features and accentuate fine details. Few attempts at image enhancement have focused on a selective approach, whereby features are enhanced based on their characteristics. This technical note presents preliminary work on a novel image enhancement algorithm based on attribute morphology. The algorithm is amorphous, that is, it does not rely on the shape of a neighbourhood operator, but rather on a feature attribute such as area or contrast. The algorithm can be directed towards differently shaped structures simultaneously without undue noise amplification. The algorithm is demonstrated through the enhancement of features in various biological images.