Super resolution method and the super resolution program which are based on hierarchical Bayes law

摘要

Plural observation pictures which photographed the identical object it offers method and the program which presume the high resolution picture on the basis of, it hides in the super resolution problem which forms the picture of a higher resolution, with Markov probability place advance distribution and the algorithm which takes in the conversion around that. You evade the problem that in addition presumption precision of parameter improve especially, in formation process of the picture, it presumes the picture which becomes dim with the technique which supposes the hierarchical advance distribution which consists of two of local smoothness and global smoothness, making use of the technique of Bayes presumption at the time of dissecting which reconstructs the field picture from the plural observation pictures which deteriorate, concerning the plural observation pictures. In addition, the hiding Markov probability place which has edge conservation is made advance distribution, identification of deterioration process of high accuracy both of picture presumption of high resolution are actualized by the fact that at the same time it converts around. Furthermore, calculation cost is made small with appropriate approximation as a polynomial order.