Sparse representation has drawn greater attention in color reconstruction. This paper proposes a color reconstruction approach based on sparse representation of classified image patches and non-local sparse coding. Firstly, image patches are classified into smooth, texture and edge classes, and different dictionaries are trained by the relevant information. Then, non-local coefficient is obtained by the calculated local coefficients. Finally, the color of the greyscale image is reconstructed by

more »

... the dictionary and the non-local coefficient. The experimental results show that the proposed method can achieve competitive performance compared with other state-of-the-art methods.