Image enhancements: deblurring and super resolution
- LecturerProf. Ming-Hsuan Yang (EECS, U. of California, Merced)
Host: Dr. Tyng-Luh Liu - Time2010-12-20 (Mon.) 10:30 – 12:00
- LocationAuditorium 106 at new IIS Building
Abstract
Abstract:
In this talk, I will present some recent results on image enhancements
by exploiting structural and sparsity information. In the first part,
I will discuss a motion deblurring algorithm that exploits sparsity
constraints of image patches using one single frame. In our
formulation, each image patch is encoded with sparse coefficients
using an over-complete dictionary. The sparsity constraints
facilitate recovering the latent image without solving an ill-posed
deconvolution problem. The dictionary is learned and updated directly
from one single frame without using additional images. The proposed
method iteratively utilizes sparsity constraints to recover latent
image, estimates the deblur kernel, and updates the dictionary
directly from one single image. The final deblurred image is then
recovered once the deblur kernel is estimated using our method. In
addition, I will present some preliminary results on finding good
features for this task.
In the second part, I will describe a super-resolution method that
exploits self-similarities and group structural information of image
patches using only one single input frame. The super-resolution
problem is posed as learning the mapping between pairs of
low-resolution and high-resolution image patches. Instead of relying
on an extrinsic set of training images as often required in
example-based super-resolution algorithms, we employ a method that
generates image pairs directly from the image pyramid of one single
frame. The generated patch pairs are clustered for training a
dictionary by enforcing group sparsity constraints underlying the
image patches. Super-resolution images are then constructed using
the learned dictionary. Furthermore, class-specific mapping can be
learned to generate high-quality images with rich texture details
and sharp edges without user assistance.
---------------------------------------------------------
Bio:
Ming-Hsuan Yang is an assistant professor in EECS at University of
California, Merced. He received the PhD degree in computer science
from University of Illinois at Urbana-Champaign in 2000. Prior to
joining UC Merced, he was a senior research scientist at Honda
Research Institute working on vision problems related to humanoid
robots. In 1999, he received the Ray Ozzie fellowship for his research
work. He coauthored the book Face Detection and Gesture Recognition
for Human-Computer Interaction (Kluwer Academic 2001) and is one of
the guest editors for a special issue on face recognition for Computer
Vision and Image Understanding, 2003 as well as a special issue on
real-world face recognition for IEEE Transactions on Pattern Analysis
and Machine Intelligence, 2011. He serves in different capacities for
a few conferences (area chair for ICCV 2011, AAAI 2011, FG 2011, ACCV
2010, CVPR 2009, ACCV 2009, CVPR 2008, and publication chair for CVPR
2010). He is an associate editor of IEEE Transactions on Pattern
Analysis and Machine Intelligence, and Image and Vision Computing.