Evolving imaging model for super-resolution reconstruction
Michał Kawulok, Paweł Benecki, Daniel Kostrzewa, Łukasz Skonieczny · Proceedings of the Genetic and Evolutionary Computation Conference Companion · 2018
Super-resolution reconstruction (SRR) allows for enhancing image spatial resolution from low-resolution (LR) observations, which are assumed to have been derived from a hypothetical high-resolution image by applying a certain imaging model (IM). However, if the actual degradation is different from the assumed IM, which is often the case in real-world scenarios, then the reconstruction quality is affected. We introduce a genetic algorithm to optimize the SRR hyper-parameters and to discover the actual IM by evolving the kernels exploited in the IM. The reported experimental results indicate that our approach outperforms the state of the art for a variety of images, including difficult real-life satellite data.