Image segmentation by multisensor data-fusion

Abstract

The information content of an image is determined by the dynamic range of the sensor and the characteristics of the signal being recorded by the sensor. The integration of information from images from different types of sensors poses several interesting subproblems that are handled in a hierarchical manner. The authors place their discussion of these problems in the context of image segmentation. Integration may take place at various levels, ranging from pixel level to logical concepts such as features, objects, and scenes. The authors give an account of their initial efforts in dealing with sensor fusion for object recognition, and they study the aspects of image segmentation in the context of sensor fusion.<>