Hardware Accelerated Super-Resolution Technology (HAST)
20120717 - 20140816:
Mr Yiu-kei Li
(1) Survey Report and Analysis Summary of existing Multi-Frame Super Resolution algorithms (2) ASTRI HAST algorithm, performance simulation models (3) System functional specification of HAST ASIC (4) Micro-architecture specifications of HAST functional modules (5) MPW test chip implementation of HAST in 65nm
Internet connectivity allows TV viewers to access a wide range of contents readily available on Internet. Market forecasts show Internet Enabled TV (IETV) has the biggest near term growth in TV market. Market leaders rush to embrace this feature in their new TV lines while all sorts of contents abound through various Internet channels. Quality of Internet delivered video contents, however, is known to be constrained by noise, low resolution and codec errors, due to bandwidth limitations. Super-Resolution is the process of obtaining higher-resolution images from one or several lower-resolution images. This project rides on the successful proliferation of Internet-enabled TV (IETV). It aims to enhance image quality for low resolution/low quality contents for large screen size displays and HDTV by Multi-frame spatial super-resolution technology. ASTRI original research on Multi-frame super-resolution algorithm will be conducted before further optimization for real time implementation hardware architecture is developed. HAST seeks to improve visual quality of any blurry, noisy or low resolution video contents to high quality and higher resolution images. It has potential applications in DVD playback, surveillance, and medical imaging enhancement beyond high-end IETV.