Image Deblurring Using Iterative Blind Deconvolution

Resource Overview

Implementation of iterative blind deconvolution for image deblurring with custom-written code avoiding MATLAB built-in functions

Detailed Documentation

During the image deblurring implementation process, I employed an iterative blind deconvolution approach and developed the entire codebase from scratch without utilizing any pre-existing MATLAB functions. This method effectively enhances image clarity and recovers fine details, resulting in more realistic and authentic visual outputs. The custom implementation involves developing an iterative algorithm that simultaneously estimates both the point spread function (PSF) and the latent sharp image through alternating optimization steps. Key components include implementing the Richardson-Lucy deconvolution variant with regularization terms to handle noise and prevent overfitting, along with developing convergence criteria to control the iteration process. The solution incorporates custom Fourier transform operations for efficient frequency domain processing and includes mechanisms for handling edge artifacts and boundary conditions.