A Preliminary Investigation On Horses Recognition Using Facial Texture Features

Horses recognition is an important task especially for horses' trainers. It is necessary in this case to identify each horse to be distinguished. All methods used for identification are invasive and threaten the well-being of horses like Tatto and Freeze branding. So, for this reason we aim to create a human method of identification using the facial biometric modality. We tested the Gabor and LBP features for face characterization and the Euclidian and Mahcosine distance for classification. We performed our approaches on our database "THoFDRL'2015 database: Horses Face Database of REGIM Lab" and we used only horses'faces in front view. These faces are considered for experimentation. The recognition rate is 95.74%. This result maintains the success of our approach in horse recognition.