A Survey on Image Enhancement with Generative Models

PhD Qualifying Examination


Title: "A Survey on Image Enhancement with Generative Models"

by

Mr. Qiang WEN


Abstract:

Image enhancement is a crucial task in the field of computer vision, with 
applications spanning domains such as medical imaging, surveillance, 
photography, and so on. The goal of image enhancement is to improve the visual 
quality and clarity of an image, making it more suitable for human perception 
and other downstream computer vision tasks such as object detection, 3D 
reconstruction and so on. Compared with traditional image enhancement 
techniques, recent advancements in generative models have introduced new and 
powerful methodologies for enhancing images.

This survey reviews the impactful literature on representative generative 
models including Variational Autoencoders (VAEs), Generative Adversarial 
Networks (GANs) and Denoising Diffusion Probabilistic Models (DDPMs). Then, we 
discuss the application of generative models to specific image enhancement 
tasks, including deblurring and super-resolution. We analyze the unique 
challenges and considerations for each task, and how the generative models have 
been adopted and extended to address these tasks effectively. Finally, this 
survey concludes by identifying emerging trends and future research directions 
in the field of image enhancement with generative models. These include the 
application of generative models in extremely challenging scenarios for image 
enhancement and the necessity of real-time generative-based image enhancement.


Date:                   Friday, 16 August 2024

Time:                   10:00am - 12:00noon

Venue:                  Room 3494
                        Lifts 25/26

Committee Members:      Dr. Dan Xu (Supervisor)
                        Dr. Qifeng Chen (Chairperson)
                        Prof. Chiew-Lan Tai
                        Prof. Nevin Zhang