SPACE-VARIANT AND SPACE-INVARIANT IMAGE DEBLURRING: A SURVEY OF HYBRID OPTIMIZATION-ENHANCED DEEP LEARNING TECHNIQUES

Abstract

An essential part of image processing is blur removal, which has vital uses in computer vision and the real world. Using a hybrid optimum deep learning model, the proposed research aims to eliminate space-variant and space-invariant blur. To boost performance and efficiency in computation, it integrates optimization methods with cutting-edge deep learning architectures. Experiments were conducted using space-variant, space-invariant, and hybrid blur types on real-world image datasets. We evaluated the suggested model's performance to that of traditional and standalone deep learning methods using measures including computational time, Structural Similarity Index (SSIM), and Peak Signal-to-Noise Ratio (PSNR). The hybrid model outperforms the existing methods since the results obtained show better mixed blur handling and superior image quality. By demonstrating a reliable scalable approach to challenging image deblurring tasks, this work advances the field of image restoration and related areas.