A method of classification for a content-based digital-image, including: defining a set of low-level features describing the semantic content of the image, the features being quantities obtainable from the image by means of logico-mathematical expressions that are known beforehand, and the choice of said features depending upon the image classes used for the classification; indexing an image to be classified, with the purpose of extracting therefrom a feature vector, the components of which consist of the values assumed, in the image, by said low-level features; splitting the feature space defined by the low-level features into a plurality of classification regions, to each one of the regions there being associated a respective image class, and each classification region being the locus of the points of the feature space defined by a finite set of conditions laid on at least one component of the feature vector; associating the feature vector to the feature space; identifying, among the classification regions, a specific classification region containing the feature vector extracted from the image to be classified; and identifying the image class associated to the specific classification region identified.
De Ponti, M., Schettini, R., Brambilla, C., Valsasna, A., Ciocca, G. (1999)Content-based digital-image classification method. . Brevetto No. T099A0996.
Content-based digital-image classification method
SCHETTINI, RAIMONDO;CIOCCA, GIANLUIGI
1999
Abstract
A method of classification for a content-based digital-image, including: defining a set of low-level features describing the semantic content of the image, the features being quantities obtainable from the image by means of logico-mathematical expressions that are known beforehand, and the choice of said features depending upon the image classes used for the classification; indexing an image to be classified, with the purpose of extracting therefrom a feature vector, the components of which consist of the values assumed, in the image, by said low-level features; splitting the feature space defined by the low-level features into a plurality of classification regions, to each one of the regions there being associated a respective image class, and each classification region being the locus of the points of the feature space defined by a finite set of conditions laid on at least one component of the feature vector; associating the feature vector to the feature space; identifying, among the classification regions, a specific classification region containing the feature vector extracted from the image to be classified; and identifying the image class associated to the specific classification region identified.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.