China
Africa
Explore chapters and articles related to this topic
Image Perfecting
Published in Leonid P. Yaroslavsky, THEORETICAL FOUNDATIONS of DIGITAL IMAGING Using MATLAB®, 2012
where λr is a set of coefficients that specify the imaging systems in the transform domain. For DFT, coefficients λr are samples of the imaging system frequency response. In ideal imaging systems, they should all be equal to unity. In reality, they usually decay with the frequency index r, which results, in particular, in image blur. Processing aimed at correcting this type of distortion is frequently referred to as image deblurring.
Integrating DeblurGAN and CNN to improve the accuracy of motion blur X-Ray image classification
Published in Journal of Nuclear Science and Technology, 2023
Ming-Chuan Chiu, Chia-Jung Wei
Image deblurring can be roughly divided into two categories: blind deblurring and non-blind deblurring. Early research focused on non-blind deblurring. Non-blind deblurring means that the blur kernel is known; for example, the Wiener filter, which reconstructs the picture by finding the minimum mean square error between the blurred image and the original image linearized manner to solve the minimization problem. However, most actual blurred images are Poisson noise, for which the recursive Richardson-Lucy algorithm was developed. The Richardson-Lucy algorithm mainly adopts Bayesian theory to iterate and compare the deblurred image with the original image to eliminate noise [30]. These mentioned research methods use deconvolution to obtain the kernel and to estimate . However, given the blur function as an unknown, scholars began to study blind deblurring algorithms [31] to estimate the sharp image and the blur kernel . The estimation process is a kind of ill-posedness because both and are unknown, and there are infinitely many solutions. This type of research typically employs image statistics and hypothetical methods to estimate the blur function. These methods solve the blurred image by considering the uniformity of the overall image blur [32]. Therefore, if one would like to restore the image, one needs meaningful information about the neighboring pixels. Many iterative-based methods have been developed to estimate the blur function through dynamic kernels and sharp images in each iteration [33].
Blind image deblurring using GLCM and negans obtuse mono proximate distance
Published in The Imaging Science Journal, 2022
Deblurring a motion blur image is a blind deconvolution task with only the blur image available to estimate the blur kernel. Deblurring becomes a non-blind deconvolution task when the blur kernel is accurately calculated. On the other hand, the deblurring issue is either blind or non-blind depending on whether the blur kernel is unknown or known. In the entire deblurring process, the estimation of the blur kernel is crucial to restoring the blurred image [2].