Abstract

We consider the dictionary learning problem for the analy-sis model based sparse representation. A novel algorithm is proposed by adapting the synthesis model based simultane-ous codeword optimisation (SimCO) algorithm to the analy-sis model. This algorithm assumes that the analysis dictio-nary contains unit 2-norm atoms and trains the dictionary by the optimisation on manifolds. This framework allows one to update multiple dictionary atoms in each iteration, lead-ing to a computationally efficient optimisation process. We demonstrate the competitive performance of the proposed al-gorithm using experiments on both synthetic and real data, as compared with three baseline algorithms, Analysis K-SVD, analysis operator learning (AOL) and learning overcomplete sparsifying transforms (LOST), respectively. Index Terms — Analysis model, SimCO, analysis dictio-nary learning.