Fingerprint image super-resolution based on multi-class deep dictionary learning and ridge prior

The identification of low-resolution fingerprints has always been one of the focuses in the field of biometric identification. This paper proposes a method for super-resolving low-resolution fingerprints based on deep dictionary learning. First, it is necessary to obtain a priori based on the fingerprint ridge orientation. After obtaining it, the ridges of the fingerprint are divided into n categories according to the direction. Each class uses deep dictionary learning models to train corresponding high- and low-resolution dictionaries respectively. In the super-resolution part, after extracting features from the patches that require super-resolution, the sparse coefficients are obtained through the deep dictionary learning model, and then combined with the high-resolution dictionary to obtain high-resolution patches, which are combined into high-resolution fingerprints. Experimental results show that the proposed method performs better than some other methods.