Motion blur is a common artifact that produces disappointing blurry images with inevitable information loss. Due to the nature of imaging sensors that accumulates incoming lights, a motion blurred image will be obtained if the camera sensor moves during exposure. Image (motion) de-blurring is a computational process to remove motion blurs from a blurred image to obtain a sharp latent image. Recently image de-blurring has become a popular topic in computer graphics and vision research, and excellent methods have been developed to improve the quality of de-blurred images and accelerate the computation speed. Image de-blurring has also a variety of applications in image enhancement software and camera industry, and a practical image de-blurring method with quality and speed would be a critical factor to improve the performance of image enhancement and camera systems.
This course will first introduce the concepts, theoretical model, problem definition, and basic approach of image de-blurring. Blind deconvolution and non-blind deconvolution are two main topics of image de-blurring, which are classified by the existence of given kernel (or PSF; point spread function) information that describes the camera motion. For both blind deconvolution and non-blind deconvolution, challenges, classical methods, and recent research trends and successful methods will be presented. A PhotoShop demo will be given to show the performance of a recently developed fast motion de-blurring method.
This course will also cover several advanced issues of image de-blurring, such as hardware based approaches, spatially-varying camera shakes, object motions, and video de-blurring. It will conclude with remaining challenges, such as outliers and noise, computation time, and quality assessment. There will be Q&A at the end of each presentation with a short discussion at the end of the course.
Level
Intermediate
Intended Audience
Researchers interested in recent advances of image deblurring and industry professionals interested in practical applications of image deblurring for camera and image restoration software
Prerequisites
General knowledge of image processing and optimization
Presenter(s)
Seungyong Lee, Pohang University of Science and Technology
Sunghyun Cho, Adobe Research
Seungyong Lee is a professor of computer science and engineering at the Pohang University of Science and Technology (POSTECH), Korea. He received the PhD degree in computer science from the Korea Advanced Institute of Science and Technology (KAIST) in 1995. From 1995 to 1996, he worked at the City College of New York as a postdoctoral research associate. Since 1996, he has been a faculty member of POSTECH, where he leads the Computer Graphics Group. From 2003 to 2004, he spent a sabbatical year at MPI Informatik in Germany as a visiting senior researcher. From 2010 to 2011, he worked at Adobe Research in Seattle as a visiting professor. His current research interests include image and video processing, non-photorealistic rendering, 3D surface reconstruction, and graphics applications.
Sunghyun Cho is a post-doctoral research scientist at Adobe Research. He received the Ph.D. in Computer Science from POSTECH in Feb. 2012, and the B.S. degree in Computer Science and in Mathematics from POSTECH in 2005. He spent six months in Beijing in 2006 working as an intern at Microsoft Research Asia. He also spent the beautiful summer and fall of 2010 in Seattle working as an intern at Adobe Research. In 2008, he was awarded Microsoft Research Asia 2008/09 Graduate Research Fellowship Award. His research interests include computational photography, image/video processing, image restoration, etc.