Computer system identifies cows by their muzzle prints
There are already several methods of identifying cattle – branding, ear tags, tattooing and ear notching all come to mind. Now, however, Egyptian scientists are working on a new biometric system that's less invasive and more difficult to thwart: electronic muzzle-printing.
Much as the whorls in our fingerprints are unique to ourselves, the ridges and valleys in the skin of each cow's muzzle (the front of their nose and mouth) is likewise unique to that animal. It's something that people have known since the 1920s, when ink prints of muzzles were used for keeping track of cattle. Since then, various studies have looked at replacing the ink and paper with high-tech tools.
The latest such technique was developed by Hamdi Mahmoud and Hagar Mohamed Reda El Hadad of Cairo's Hamdi Mahmoud of BeniSuef University. In their setup, close-up images of cow muzzles are processed utilizing a machine learning system known as a multiclass support vector machine (MSVM). Among other things, support vector machines are known for their ability to classify items based on pattern recognition.
In real-world tests, the MSVM has so far been able to identify cows based on their muzzle prints with 94 percent accuracy. The scientists are now working on improving that figure, along with speeding up the processing time.
A paper on the research was recently published in the International Journal of Image Mining.
Source: Inderscience Publishers