Guorong Yu presents Image registration based on probabilistic graph model and multi-spectral image registration

On 2018-12-20 11:00:00 at G205, Karlovo náměstí 13, Praha 2
This talk has two parts, including my former work and current work. My former
work is about image matching based on probabilistic graph model. My current
is about multi spectral image registration. Accurate image registration is a
vital step in many computer vision processes. However, traditional SIFT based
methods are not able to obtain satisfactory results in some cases. We turn the
matching problem into a Markov Random Field (MRF) optimization problem and
propose a new image registration method based on message passing to screen out
mismatches caused by the conventional SIFT-based algorithm. The proposed method
comprises three steps. First, a SIFT detector is used to detect key points and
then these points are described by Daisy instead of the SIFT detector (to
improve efficiency); second, a new geometric constraint with distance and
direction terms is designed and incorporated into sequential tree-reweighted
message passing (TRW-S) – this obtains the coarse matching result; finally,
the RANSAC algorithm is adopted to filter out the mismatches and obtain an
accurate transformation. The experimental results demonstrate that the proposed
method significantly improves the efficiency and the accuracy of the
registration process and secures a desirable result.
My current work is about multi spectral image registration. Multi spectral
images are captured by different sensors and there exist non linear intensity
changes between them. Conventional gradient based method such as SIFT cannot
tackle this problem. We convolve images with log gabor filters and obtain phase
congruency from the filter response. Then these phase congruency are utilized
generate descriptors for each keypoint. Preliminary experiments demonstrate
this descriptor outperforms the SIFT descriptor.
Guorong Yu is a visiting PhD student from Wuhan University. He will be working
with Jan Kybic. The purpose of this talk is to introduce him and his work. The
first part of this talk is based on his article:

Shuangming Zhao, Guorong Yu: A new image registration algorithm using SDTR.
Neurocomputing Volume 234, 19 April 2017, Pages 174-184
Responsible person: Petr Pošík