MAP-Based Image-Analogies Super-Resolution Reconstruction
Zhou Ming-zheng · Computer Technology and Development · 2009
Presents a new method(Maximum A Posterior Image Analogies,MAPIA) based on MAP for image super resolution in use of image analogies.Firstly,use the Bayesian MAP algorithm for supper resolution image reconstruction;secondly,use a low resolution image and it's super-resolution version to construct the training set.Learn the relationship between source image pair,and then,fit this problem to image analogies for super resolution.The experiments show the approach can amplify images,and the edges in the MAPIA reconstruction result are much clearner and the details are more prominent over traditional methods.