Super-resolution Image Reconstruction means constructing a high-resolution image or more high-resolution images from a sequence of low resolution images.Firstly, this paper provided an overview of the mathematic principle of Super-resolution Image Construction and introduced several common algorithms. Then, according to the building of imaging model, it was found that Super-resolution Image Reconstruction was an ill-posed inverse problem. Such problems could be computed by adding prior knowledge which constrained the feasible solution space. Finally, solution space was constrained by exploring Markov Random Field(MRF), based on the MAP frame model. Meanwhile, through using Gibbs Random Field(GRF), limitation of the MRF was broken, so redundant information between image pixels was made explicit use of to construct high-resolution images from local region. Compared with traditional image expansion, this not only improved the resolution, but also preserved more detail information. |