Real-time
BASIS Feature Descriptor Applications
|
|
We used our feature
descriptors for two applications. The
first one is for pose estimation for UAVs and
the other one is for book spine feature matching
for library shelf reading application.
Feature point
matching is an important step for many
vision-based UAV applications. We used our
feature descriptors for feature point matching
that is well suited for micro-UAVs equipped with
a low-resource, compact, lightweight, and
low-power embedded vision sensor. The
BAsis Sparse-coding Inspired Similarity (BASIS)
descriptor requires simple mathematical
operations, uses comparatively small memory
storage, and can support color and grayscale
feature description. It is an excellent
candidate for implementation on low-resource
systems that require real-time performance,
where complex mathematical operations are
prohibitively expensive. To demonstrate its
performance, the feature matching result was
used to calculate a frame-to-frame homography
that is essential to UAV applications such as
pose estimation and obstacle detection for
navigation.
The library book
inventory or so called shelf-reading process is
an important task for libraries to identify
missing books and to locate misplaced
books. Currently, this process requires
humans to remove each book from the shelf, open
the front cover, scan its barcode, and reshelve
the book. It is a labor intensive and often
error-prone process. A vision-based automation
system is proposed to improve this process
without those prohibitively high upfront
costs. This low-cost shelf-reading system
includes a hand-held imaging device such as a
smartphone with wireless communication
capability and a server that processes the
captured book spine images for book
identification. Existing color feature
descriptors for feature matching typically use
gray scale feature detectors, which omit
important color edges. Also, photometric
invariant color feature descriptors require
complex computations that are not necessary to
provide color descriptor information.
|
Graduate Students:
|
Spencer Fowers
|
Publications:
- S.G. Fowers and D.J. Lee,
“An Effective Color Addition to Feature
Detection and Description for Book Spine Image
Matching,” International Scholarly
Research Network–Machine Vision, vol. 2012,
Article ID 945973, 15 pages, January 2012.
- S.G.
Fowers, D.J. Lee, and G.M. Xiong, “Improve
Library Shelf Reading Using Color Feature
Matching of Book-Spine Image,”
IEEE Proceedings of the International Conference
on Control, Automation, Robotics and Vision
(ICARCV) , p. 2160-2165, Singapore, December
7-10, 2010.
|
(Click
image to view.)
|
|
|