Abstract
Recognition of cattle based on muzzle point image pattern (nose print) is a well study problem in the field of animal biometrics, computer vision, pattern recognition and various application domains. Missed cattle, false insurance claims and relocation at slaughter houses are major problems throughout the world. Muzzle pattern of cattle is a suitable biometric trait to recognize them by extracted features from muzzle pattern by using computer vision and pattern recognition approaches. It is similar to human's fingerprint recognition. However, the accuracy of animal biometric recognition systems is affected due to problems of low illumination condition, pose and recognition of animal at given distance. Feature selection is known to be a critical step in the design of pattern recognition and classifier for several reasons. It selects a discriminant feature vector set or pre-specified number of features from muzzle pattern database that leads to the best possible performance of the entire classifier in muzzle recognition of cattle. This book chapter presents a novel method of feature selection by using Hybrid Chaos Particle Swarm Optimization (PSO) and Bacterial Foraging Optimization (BFO) techniques. It has two parts: first, two types of chaotic mappings are introduced in different phase of hybrid algorithms which preserve the diversity of population and improve the global searching capability; (2) this book chapter exploited holistic feature approaches: Principal Component Analysis (PCA), Local Discriminant Analysis (LDA) and Discrete Cosine Transform (DCT) [28, 85] extract feature from the muzzle pattern images of cattle. Then, feature (eigenvector), fisher face and DCT feature vector are selected by applying hybrid PSO and BFO metaheuristic approach; it quickly find out the subspace of feature that is most beneficial to classification and recognition of muzzle pattern of cattle. This chapter provides with the stepping stone for future researches to unveil how swarm intelligence algorithms can solve the complex optimization problems and feature selection with helps to improve the cattle identification accuracy.