Image super-resolution by parameter multisurface fitting

In this paper, we propose a novel super-resolution algorithm by using parameter multi-surface fitting. In order to utilize the spatial information in a more effective way, we use the unknown high-resolution pixels as a parameter of fitted surface in the multi-surface fitting equations. We form one surface at each low-resolution pixel in the neighborhood of each high-resolution pixel. The surface is formed based on the information from both the known LR pixels and unknown HR pixel which is used as a parameter. The final result is obtained by fusing the sampling values from these surfaces in the maximum a posteriori fashion. The proposed method can reduce the fitting errors of the surfaces and assign different weights to the sampling values. Experimental results show the superiority of the proposed method comparing to the state-of-arts.