Cattle Identification from Muzzle Print Image Pattern Using Hybrid Feature Descriptors and SVM

Abstract

Biometric-based recognition is gaining popularity based on its applications and uses. Identification system in animals is applied on the biometric features, visual features, phenotype traits, and genotype traits. The muzzle point image pattern is a primary biometric feature for the identification of the individual cattle. This feature is very similar to minutiae points of human fingerprints. The study presents the individual cattle identification based on the muzzle image pattern of cattle. The identification is required for the verification of false insurance claims, registration of cattle, and traceability of natural habitat of cattle. The proposed system uses the texture feature descriptor, scale-invariant, and speeded up robust to extract features from the muzzle point images pattern. By using muzzle image pattern database of 930 images, this proposed algorithm yields the desired results of 90.22% identification accuracy.