Novel Super-resolution algorithm

2015 Fiscal Year Final Research Report

• Project/Area Number: 23560451
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Communication/Network engineering
• Research Institution: Nagoya Institute of Technology
• Principal Investigator: Sakurai Masaru 名古屋工業大学, 工学(系)研究科(研究院), 教授 (70432284)
• Co-Investigator(Kenkyū-buntansha): GOTO Tomio 名古屋工業大学, 情報工学専攻, 助教 (20324478)
• Project Period (FY): 2011-04-28 – 2016-03-31
• Keywords: 超解像 / TV正則化フィルタ / Shock フィルタ / 事例学習法

Outline of Final Research Achievements

We have developed a super-resolution system that consits of TV(Total Variation) reguralization filter and Shock filter. The input image is devided into the structure component and the texture component. The structure component is enhanced by the shock filter, and the texture component is enhanced by the pulse enhancement filter. The result is much better than the conventuional method. This system is implemented on the GPU. The conputational time is less than 16.7msec that is 1 frame time of the television. This enable that the system can be implemented on 4K-HDTV. Its also applicable to the medical image.In the future, since FPGA implementation will be required, we study FPGA implementation of this system.

Free Research Field

情報学