An Efficient Blind Approach for Restoration of Atmospheric Turbulence Degraded Images

Image restoration is a technique to improve the quality of degraded image using the estimated amount of noise and blur involved in the image. The image may get degraded due to different atmospheric, environmental conditions, transmission medium or faulty devices. This paper focuses on restoration of atmospheric turbulence affected sequences. Lots of image registration techniques are used for turbulence mitigation which causes smoothness. This paper proposes a novel blind deconvolution approach. The proposed method includes central moments to estimate the Point Spread Function (PSF) blindly. The proposed method is computed on standard dataset (OTIS) and is compared with state of the art Adaptive kurtosis and kurtosis minimization approach. Experimental results showcases that the proposed approach achieves higher accuracy on standard parameters like PSNR and SSIM.