Since the advent of smart phones, the act of capturing images has become ingrained in human behavior, evolving into an integral aspect of daily life. The human perception of image quality is a crucial area of research as individuals acquire and interpret images regularly. Automation in image processing systems plays a pivotal role in quantifying image quality, seamlessly addressing factors like blurriness, noise, and compression that can significantly degrade the visual experience. While image processing professionals can promptly identify such distortions, the incorporation of human visual perception into automated system design is often challenging. Consequently, the evaluation of image quality by a machine stands as a critical research domain. Image quality assessment algorithms are employed to gauge the quality of images, with the objective of approximating human visual judgments. Although deep learning has demonstrated remarkable success in addressing various real-world challenges, particularly in tasks like object detection and image classification, its application in image quality assessment is still evolving. The primary challenge lies in assessing image quality without explicit consideration of human perception, presenting a dynamic area of ongoing research. A key hurdle in this research is the scarcity of images accompanied by quality scores, specifically differential average feedback scores provided by image quality experts on a 1-100 scale. This limitation hinders the development of robust models for image quality assessment, emphasizing the need for innovative solutions in this dynamic research domain.

Image quality assessment, Human visual perception, Image processing systems, Automation, Deep learning, Visual distortions, Quality scores, Differential feedback