Context categorization is a fundamental pre-requisite for multi-domainmultimedia content analysis applications in order to manage contextualinformation in an efficient manner. In this paper, we introduce a new colorimage context categorization method (DITEC) based on the trace transform. Theproblem of dimensionality reduction of the obtained trace transform signal isaddressed through statistical descriptors that keep the underlying information.These extracted features offer a highly discriminant behavior for contentcategorization. The theoretical properties of the method are analyzed andvalidated experimentally through two different datasets.
展开▼
