Scanner Helps Track Zebras’ Bar Code
Zebras – those living,
breathing bar codes – can now be scanned by a method that could help scientists
track individual animals in the herd and add more black-and-white facts to our knowledge
of the animal kingdom.
The StripeSpotter automatic identification system for individual
animals with prominent stripes or patches was developed collaboratively by scientists
at Princeton University in New Jersey and at the University of Illinois at Chicago.
Two species of zebra in Kenya, Africa, were the (sometimes unwilling) participants
in the study.
While doing field research, the scientists captured digital images
of the zebras in the wild, then cropped the images to a region of interest –
an area of the animal’s anatomy that appeared consistent across a number of
pictures. Using their own StripeCodes algorithm and “noisy” photographs
taken in the wild, they built a biometric database of individual animals, differentiated
by their coat markings. The “noisy” aspects of the photos included uncooperative subjects, coat deformities, and variations in image exposure, scale and perspective.
The investigators compared the coat markings in a new image of
an unknown animal against the database to determine whether the animal had been
previously observed and identified with the system. The algorithm extracted simple
image features and used a dynamic programming algorithm to compare images.
A database of digital zebra images and an algorithmic approach help
scientists track individual zebras in the wild. The images are cropped
to the same
region of interest in the animal’s anatomy. The image of an unidentified
animal’s coat markings are compared against those in the database to
determine if the animal has been previously sighted. Photos courtesy of
Identifying individual wild animals as they are sighted going
about their daily lives has been a challenge. Embedding an electronic tracking device
in a wild animal can be risky business, not to mention expensive. Other options,
such as manually recording data on animals from photographs or video, obtaining
genetic markings from excrement, or capture-and-recapture techniques, have been
problematic in terms of cost, efficiency or accuracy.
As for these “markedly” exotic creatures, hopefully
the saying is true – a leopard cannot change its spots, nor a zebra its stripes
– as we set out to identify them by those markings.
MORE FROM PHOTONICS MEDIA