Barcode Scanning for Zebras
You’ve more than likely seen a Zebra before, but did you ever make the connection that their stripes somewhat resemble a barcode? Now, a team of US biologists and computer scientists have come up with a “zebra scanner,” which is able to identify individual animals based on a still photo.
Their “zebra scanner,” coined StripeSpotter, requires only a small amount of human input. By using a photograph of an individual zebra, scientists can select part of the zebra’s side. This visual information is then automatically sliced into a number of horizontal bands and each pixel is transformed into 100% black or white, creating a high-contrast version of the zebra’s stripes. From there, each stripe is encoded as a “StripeString,” which is basically a sequence of colored blocks that represent different lengths. This collection of StripeStrings forms a SripeCode, the zebra equivalent of a barcode.
Once a zebra has been entered into the database and given a StripeCode, the researchers match another picture of the same animal by comparing the StripeStrings of the new and original images. From there, each image generates a different set of StripeStrings, but the original ratios of white to black remain similar. The StripeCode with the most similar StripeStrings in the database will match to identify the correct animal. The researchers plan to present their work at the International Conference on Multimedia Retrievalin in Italy later this month.
Zebras have always reminded me of barcodes, and evidently our friends at Zebra Technologies felt the same way! Let us know what you think about StripeSpotter, the barcode system for zebras, by commenting here, or on our facebook or twitter pages.