IMAGE PARTITION TRANSFORMS FOR FAITHFUL SEGMENTATION SEARCH
Dmitry Kinoshenko, Sergii Mashtalir, Konstantin Shcherbinin, Elena Yegorova · 2007
The explosion of image content is closely connected with segmentations efficiency. High-level region- based interpretations are associated with some a priori information, measurable region properties, heuristics, and plausibility of computational inference. Conventional similarity analysis consists of following steps: images are segmented into disjoint regions, features are extracted from each region and the set of all features is used for high-level processing. Quite often simultaneous processing of several partitions is desired in order to produce reliable true conclusion. We propose operations with segmented images and a metric for nested partitions.