Animal identification based on footprint recognition

Abstract

Animals can be identified using their footprints. Several features contained within an animal footprint can be used to aid in the identification of an animal. Amongst these features, the most common and most used by humans to manually identify the animal is the number and size of blobs in the footprint. Using image processing techniques an algorithm was created to segment and extract the best possible representation of the footprint which varied across color. Connected Components was then used to count the number of blobs contained within the footprint and measure the size of each blob. Using this information alone, it was found that a footprint could accurately be classified as either hoofed, padded or full print. Finally morphological feature extraction techniques were investigated to fully classify the footprint. The system implemented boasted a 97% accuracy rate.