Improved local minimum pixel prior for blind restoration algorithm of remote sensing images
In order to solve the problem of inaccurate fuzzy kernel estimation and ringing effect in restored images during blind restoration of remote sensing images,an improved local minimum pixel a priori remote sensing image blind restoration algorithm is proposed.The algorithm introduces the combination of extreme channel a priori and local minimum pixel a priori to better constrain the intensity of the image,which is conducive to obtaining a better poten-tially clear image;then the gradient-based method is used to estimate the fuzzy kernel,and the fuzzy kernel estima-tion is carried out alternatively and iteratively with the estimation of the intermediate potentially clear image to ob-tain a more desirable fuzzy kernel.Finally,an improved Laplace and regularized image restoration algorithm is used to input the resulting fuzzy kernel.That is,a joint bilateral filter is introduced to suppress the ringing effect of im-age restoration.The experimental results show that the method in this paper has a good effect on remote sensing im-age restoration,the restored image has clear edges,the ringing artifacts are suppressed and the fuzzy kernel is more ideal.The objective evaluation index the peak signal to noise ratio(PSNR)is improved by about 1.40 dB and the structural similarity(SSIM)is increased by about 0.02 on average compared with the cutting-edge restoration algo-rithm.
万方数据
维普
