A Human Face Super-resolution Reconstruction Algorithm Based on Niche Genetic Algorithm

A super-resolution restoration algorithm based on sample learning is to implement the “learning” between the images with high resolution and low resolution usually in the way of “search and paste”. To improve the quality of restored image by building a huge sample database. The mismatching in search results may lead lower reconstructed quality. To solve these questions, this paper proposes a facial image super-resolution reconstruction algorithm based on niche genetic algorithm. The algorithm processes in regions based on facial image’s textural features, to divide the solve space into multi layers, different distance constants can be used in each layer and to take advantage of the features in a niche that only have one individual to ensure algorithm would not easy to be involved in local optimum. The experimental results show that our algorithm can reach a preferable reconstructed quality and with faster running speed, which is an effective facial image super-resolution reconstruction method.