Flexible architectures for retinal blood vessel segmentation in high-resolution fundus images

Flexible architectures for retinal blood vessel segmentation in high-resolution fundus images Blood vessel segmentation from high-resolution fundus images is a necessary step in several retinal pathologies detection. Automatic blood vessel segmentation is a computing-intensive task, which raises the need for acceleration with hardware architectures. In this paper, we propose two architectures for blood vessel segmentation using a matched filter (MF). The first architecture is a scalable hardware architecture, while the second one is an application-specific instruction-set processor. An efficient, real-time hardware implementation of the algorithm is made possible through parallel processing and efficient resource sharing. A tool for the automatic generation of particularized HDL descriptions of the architecture is proposed. The tool starts from a common architecture template and takes as input the parameters of the MF. A designer thus gains a significant amount of flexibility and productivity with the parameter selection problem and the evaluation of corresponding implementations. Several designs were verified and implemented on an FPGA platform. Performance in terms of area utilization and maximum frequency are reported. The results show significant improvement over state-of-the-art implementations, by up to a factor of 14× for high-resolution fundus images. The second architecture is based on the Tensilica Xtensa LX processor. With only two additional custom instructions requiring an additional 4× the area of the basic processor, the ASIP achieves a significant speedup of 7.76× when compared to the basic processor, while retaining all its flexibility. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Real-Time Image Processing Springer Journals

Flexible architectures for retinal blood vessel segmentation in high-resolution fundus images

Loading next page...
 
/lp/springer_journal/flexible-architectures-for-retinal-blood-vessel-segmentation-in-high-L51uGMULMI
Publisher
Springer Berlin Heidelberg
Copyright
Copyright © 2016 by Springer-Verlag Berlin Heidelberg
Subject
Computer Science; Image Processing and Computer Vision; Multimedia Information Systems; Computer Graphics; Pattern Recognition; Signal,Image and Speech Processing
ISSN
1861-8200
eISSN
1861-8219
D.O.I.
10.1007/s11554-016-0661-4
Publisher site
See Article on Publisher Site

Abstract

Blood vessel segmentation from high-resolution fundus images is a necessary step in several retinal pathologies detection. Automatic blood vessel segmentation is a computing-intensive task, which raises the need for acceleration with hardware architectures. In this paper, we propose two architectures for blood vessel segmentation using a matched filter (MF). The first architecture is a scalable hardware architecture, while the second one is an application-specific instruction-set processor. An efficient, real-time hardware implementation of the algorithm is made possible through parallel processing and efficient resource sharing. A tool for the automatic generation of particularized HDL descriptions of the architecture is proposed. The tool starts from a common architecture template and takes as input the parameters of the MF. A designer thus gains a significant amount of flexibility and productivity with the parameter selection problem and the evaluation of corresponding implementations. Several designs were verified and implemented on an FPGA platform. Performance in terms of area utilization and maximum frequency are reported. The results show significant improvement over state-of-the-art implementations, by up to a factor of 14× for high-resolution fundus images. The second architecture is based on the Tensilica Xtensa LX processor. With only two additional custom instructions requiring an additional 4× the area of the basic processor, the ASIP achieves a significant speedup of 7.76× when compared to the basic processor, while retaining all its flexibility.

Journal

Journal of Real-Time Image ProcessingSpringer Journals

Published: Dec 20, 2016

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off