Blind multi-image super resolution reconstruction with Gaussian blur and Gaussian noise
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(4) 68-73
Institute of Computer Science and Technology, Yibin University, Wuliangye Str.8, Yibin, China
A framework of blind multi-image super resolution reconstruction method is proposed to improve the resolution of low resolution images with Gaussian blur and noise. In the low resolution imaging model, the shift motion, Gaussian blur, down-sampling, as well as Gaussian noise are all considered. Firstly, the Gaussian noise in the low resolution image is reduced through Wiener filtering method. Secondly, the Gaussian blur of the de-noised image is estimated through error-parameter analysis method. Thirdly, the motion parameters are estimated. Finally, super resolution reconstruction is performed through iterative back projection algorithm. Experimental results show that the Gaussian blur and motion parameters are estimated with high precision, and that the Gaussian noise is restrained effectively. The visual effect and peak signal to noise ratio (PSNR) of the super resolution reconstructed image are enhanced. The importance of Gaussian blur estimation and effect of Gaussian de-noising in multi-image super resolution reconstruction are tested in an experimental way.