2027-01-18 Technical Sharing Session – AI-Based Image Restoration for Under-Display Cameras: A Four-Stage Pipeline Under Power and Latency Constraints

Organized by HKIE Electronics Division

Date, Time & Venue
Date: 2027-01-18
Time: 18:30-20:00
Venue: Chan Yat Mei Sophie Room, 9/F, HKIE Headquarters, Island Beverley, No. 1 Great George Street, Causeway Bay, Hong Kong

Programme highlights

As displays evolve, AI-based image processing is critical to solving the optical limitations of Under-Display Cameras (UDCs). This seminar provides a comprehensive overview of the four core algorithmic stages in UDC restoration: light-transmission gain compensation, deconvolution for optical refraction and blur, multi-frame temporal noise reduction and alignment, and white-balance correction for color shifts. Focusing on practical constraints like power consumption and latency, the discussion examines state-of-the-art AI-based methods, comparing them against conventional techniques across each stage. Participants will explore representative AI approaches, analyze key trade-offs, including benefits, limitations, and resource demands, and gain practical insights into the open challenges of deploying real-time AI image restoration on commercial devices.

Speaker

Dr. Hon-Tung Luk, Senior Engineer in the AI & Data Science team at HKPC, is an expert in computer vision and artificial intelligence with over a decade of research and industrial experience. Holding a Ph.D. in Information Engineering and dual bachelor’s degrees in Mathematics and Information Engineering from CUHK, his career spans key technical roles including Senior Computer Vision Engineer at TCL Corporate Research. His work encompasses advanced industrial computer vision, image quality enhancement, vision-language models, and enterprise LLM applications such as multi-agent systems and retrieval-augmented generation (RAG). The holder of multiple US and China patents and author of publications in top venues like IEEE Transactions, Dr. Luk seamlessly bridges academic rigor with practical engineering to advance real-world AI deployment.

Registration & enquiries

The seminar is free of charge with maximum of participants of 70. For registration, please complete the online Enrolment Form at END website Here. Successful applicants will be notified before the event. For enquiries, please contact Nicole WONG at 6547-1344. Attendant certificate will be awarded after seminar.

Skip to content