Image resolution enhancement using dual-tree complex wavelet transform and neural networks
Shirin Salehi, Homayoun Mahdavi‐Nasab, Hossein Pourghassem · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2011
In this paper, an image resolution enhancement algorithm using dual-tree complex wavelet transform and neural network is proposed. The low resolution image is used as the input signal of the neural network and the network estimates all the 16 wavelet subbands of the corresponding high resolution image. The high resolution image is then obtained using the inverse dual-tree complex wavelet transform. Directional selectivity of the DT-CWT increases the accuracy of subband estimation done by the neural network. Subjective comparisons and objective quality assessments indicate additional improvement over the conventional bicubic and bilinear interpolation techniques and even new complex counterpart methods.