Semi Supervised Feature Extraction for Filling Semantic Gap in ImageRetrieval

In this paper, a novel framework for combining thetexture, shape information, beside that newly introduced transformfor textural features are presented. This method is based on SpectralFunction that provides a statistical description in the frequencydomain of signals, and then the Spectal function (SF) of each signalis calculated by spectral analyzer (SSA). Features are energy andstandard deviation of SF of signals got at different regions ofbifrequency plane. This scheme shows high performance in Imagesets. The experimental results are compared with previous works andare found to be encouraging

کلیدواژه ها:

Spectral Function ، Spectral Analysis ، Localdescriptors ، Gradient vector