Progressive image reconstruction via cellular neural networks

Abstract

The progressive image reconstruction via CNN is presented, where the CNN template mapping an image to a domain concerned with the coefficients of the radial basis function network for image interpolation is provided. The analog CNN dynamics achieves massively parallel computing, thus, the proposed procedure would create a new paradigm of CNN at the point of very high-speed image decoding and encoding. The simulation results shows good performance of the proposed CNN for image reconstruction.