A Spectrum Based Algorithm for Image Classification

In this paper, a novel algorithm for imageclassification is presented which uses the projective value ofadjacency spectrum as classified samples. Firstly, the eigenvalues ofadjacency matrices constructed on the feature point-sets of images areobtained by singular value decomposition. Secondly, the eigenvalues areprojected onto the eigenspace by means of the covariance matrix.Finally, image classification is performed by adopting RBF and PNNneural networks as classifiers respectively. Meanwhile, sometheoretical analyses are given to support the proposed method.