Shape description for content-based image retrieval

The present work is focused on a global image characterization based on a description of the 2D displacements of the different shapes present in the image, which can be employed for CBIR. applications. To this aim, a recognition system has been developed, that detects automatically image ROIs containing single objects, and classifies them as belonging to a particular class of shapes. In our approach we make use of the eigenvalues of the covariance matrix computed from the pixel rows of a single ROI These quantities are arranged in a vector form, and are classified using Support Vector Machines (SVMs). The selected feature allows us to recognize shapes in a robust fashion, despite rotations or scaling, and, to some extent, independently from the light conditions. Theoretical foundations of the approach are presented in the paper, together with an outline of the system, and some preliminary experimental results.