Image recognition, feature extraction, feature vector, standard deviation, statistical dispersion, unsupervised classification, automatic clustering, region-growing.
Our paper focuses on the graphical analysis domain. We propose an automatic image recognition technique. This approach consists of two main pattern recognition steps. First, it performs an image feature extraction operation on an input image set, using statistical dispersion features. Then, an unsupervised classification process is performed on the previously obtained graphical feature vectors. An automatic region-growing based clustering procedure is proposed and utilized in the classification stage.
Institute for Computer Sciences
Romanian Academy, Iasi branch
– 0.16 Mb