Deep Back-Projection Networks For Super-Resolution

The feed-forward architectures of recently proposed deep super-resolution networks learn representations of low-resolution inputs, and the non-linear mapping from those to high-resolution output. However, this approach does not fully address the mutual dependencies of lowand high-resolution images. We propose Deep Back-Projection Networks (DBPN), that exploit iterative upand downsampling layers, providing an error feedback mechanism for projection errors at each stage. We construct mutuallyconnected upand down-sampling stages each of which represents different types of image degradation and highresolution components. We show that extending this idea to allow concatenation of features across upand downsampling stages (Dense DBPN) allows us to reconstruct further improve super-resolution, yielding superior results and in particular establishing new state of the art results for large scaling factors such as 8× across multiple data sets.