Online update joint dictionary learning method for hyperspectral image super resolution

Abstract

Aiming at solving the problems that the commonly used single-dictionary learning and offline-update dictionary learning are not precise enough for the sparse coding of input image signal, the quality of reconstruction is not high and the robustness is poor in current image super-resolution reconstruction, online update joint dictionary learning method is proposed. According to the reconstruction constraints and sparse prior, the least square method is used to learn and update the joint dictionary, which can effectively improve the precision of sparse coding. Experimental results show that the proposed method can improve the quality and robustness of reconstructed hyperspectral images when compared with the classical image super-resolution reconstruction algorithm, ensure that the computation time does not increase excessively.