Semi Supervised Feature Extraction for Filling Semantic Gap in Image Retrieval
Mahdi Jalali, Tohid Sedghi · 2011
In this paper, a novel framework for combining the texture, shape information, beside that newly introduced transform for textural features are presented. This method is based on Spectral Function that provides a statistical description in the frequency domain of signals, and then the Spectal function (SF) of each signal is calculated by spectral analyzer (SSA). Features are energy and standard deviation of SF of signals got at different regions of bifrequency plane. This scheme shows high performance in Image sets. The experimental results are compared with previous works and are found to be encouraging.