IMAGE ANALYSIS AS CONTEXT-BASED REASONING

Abstract

This paper proposes a new methodology that treats image analysis problems as determining a sequence of spatial contexts by reasoning about a sequence of reference objects and their spatial relationships with other objects. The approach has been experimented with medical image analysis and interpretation. The result has demonstrated that reference objects in image analysis is a very effective strategy computer image analysis systems that can purposely focuses the image analysis effort on the most promising part of the entire image to localize the target object quickly and accurately by eliminating most potential false positives.