Super-resolution image processing with total variation regularization and shock filters
Masashi Watanabe, Yasutaka Sakuta, Tomio Goto, Satoshi Hirano, Masaru Sakurai · 2012
In this paper, we propose a new super-resolution system based on the total variation (TV) regularization method. The TV regularization method is known to be a very effective method for image magnification with edge enhancement. A problem with the conventional TV magnification method is its high computational time requirement. We propose the introduction of a shock filter in the conventional TV system. Our results show good performance in terms of both better edge enhancement and much lower computational time, as compared with the conventional method. We implemented our system using a multi-core CELL processor, and we verified the possibility of applying our system to an HDTV receiver.