Muzzle Based Identification of Cattle Using KAZE

Biometric Identification for animals has been an emerging research field in computer vision. Biometric Identification plays an important role in monitoring diseases, vaccination, planning and control of the produce, and also in ownership assignment. There are several Traditional identification methods like the Ear-Tagging, Ear-Notching, Ear-Tattooing, Freeze-Branding, Hot-Branding and Electrical methods using RFID. The Traditional methods have been invasive, easily duplicable. They are also known for their low accuracies in identification as they are vulnerable to losses. A system with better performance is much needed in this field. Visual Animal Biometrics is gaining wide acceptance all over the world as it provides with better results. This paper aims to explain in detail the implementation of a feature extraction technique called KAZE and through experimental analysis show that it performs better than other feature extraction algorithms.