Authors
Nafikov M.
|
Degree
|
Chelyabinsk State University |
|---|---|
|
E-mail
|
nafikov.maksim.emp@gmail.com |
|
Location
|
Chelyabinsk |
|
Articles
|
Automatic detection of blood vessels in retinal images using median filter and Gabor filtersNowadays the 3d printing lets us produce cheap retina scanners. Researchers have printed an ophthalmoscope
adapter for a smartphone to observe a retina using the smartphone camera. Thus, the main
disadvantage of the retina scanners disappears; moreover, the invention lets us use retina based biometric
authentication and automatic disease detection systems using only the smartphone. Both the biometric
authentication systems and the automatic disease detection systems require features characterizing
the identity or the disease. Such features are blood vessels and this article is devoted to detection
of them. Two algorithms of the blood vessels segmentation based on median filter and Gabor filters are
developed and represented. Steps of preliminary image processing also are described. The first step is
contrast enhancement based on using green channel and contrast limited adaptive histogram equalization,
and the second step is background exclusion based on high frequency filter. The first segmentation
algorithm uses automatic thresholding Otsu, median filter and filter by length. The second algorithm
uses Gabor filters and analyzing histogram of the image applies automatic thresholding. To estimate
the performance of the proposed algorithms, tests have been conducted on the two databases:
DRIVE and STARE. The results show that the segmentation algorithm based on median filter can be
applied both in the biometric authentication systems and in the automatic disease detection systems.
The second segmentation algorithm requires lots of computing power, and therefore can’t be applied
in the biometric authentication systems. The 3d printing technology lets us use the retina scanners by
almost any smartphones and further researchers in development of the retina based automatic disease
detection systems provide detecting of the blood vessels disease in the early stages and monitoring progression
of the disease only by our smartphones. It will undoubtedly improve the level of healthcare.
Project materials are available at https://github.com/forcesh/authentication_based_on_retinal_images
Read more...
|
