On measuring low-level self and relative saliency in photographic images

Measuring perceptual saliency of different regions in a scene is important for determining regions of interest (ROI). Color, texture and shape cues ar e good low-level features for detecting saliency. While self saliency refer s to intrinsic attributes of a particular region, relative saliency is used to measure how salient a region is relative to its surrounding and thus ne eds to be defined within a spatial context. A few spatial context models ar e investigated in this study. In particular, we propose an auto-scaled, ext ended neighborhood-based context model to obtain reliable measurements of r elative saliency features. Comparison of three context models has shown tha t the proposed model is capable of generating predicates more consistent wi th perceived saliency. (C) 2001 Elsevier Science B.V. All rights reserved.