A Novel Method for Video Retrieval Using Relevant Feedback
Yongbin Shen, Yixin Yan, Shengming Li · 2009
In this paper, we propose a relevant feedback algorithm based on supervised classification and unsupervised feature extraction using independent component analysis (ICA). First, we calculate the approximate coefficients of video frames in a single shot by two-level Haar wavelet decomposition; second,ICA is utilized to obtain the independent components of the wavelet coefficients, which are used to represent video frames. Simulation results demonstrated the validity of the algorithm.