A NOVEL SCHEME FOR DENOISING OF RGB IMAGES USING GENETIC ALGORITHM
Amit Jain, Manpreet Singh · Journal of Critical Reviews · 2020
Abstract: Image Deblurring is an important image enhancement tool required by both image processing applications for artists as well as for computer vision systems. Many different techniques for deblurring of a blurred image have been developed. In this paper, we are presenting a new technique for image deblurring with the application of optimization algorithm called genetic algorithm. The work uses GA to refine the Kernel function used for deblurring of Image. The kernel for blurred is image is computed first by using Laplacianof Gaussian (LoG) filtering and then applying morphological operations to extract the PSF. The PSF or kernel function thus computed is then used to compute Mean Squared Error when applied to deblur the image. The resultant MSE is utilized as objective function and is minimized by adjusting the parameters for LoGfilter and blurring filters in term of chromosomes in Genetic Algorithm. After successful iterations, the MSE is minimized and best fit kernel function used to deblur the image. Different quantitative performance metrics like MSE, Coefficient of Correlation and Standard Deviation etc are used to evaluate our work. The result of our technique is compared with existing blind deblurringalgorithm