A real-time computer vision monitoring way for animal diversity

It's laborious and time consuming for manual measuring of animal diversity in city's greenbelt. It's also impossible for people to perform continuous and all-weather measuring by manual way. Therefore, the objective of this endeavor was to explore a real-time computer-vision system that allows continuous measuring of animal diversity. The system was developed to perform image processing algorithms to get the animal images and extract their features. Following moment detection, firstly, the fuzzy c-means clustering was applied to color image segmentation. Secondly, morphological filtering was performed to eliminate noises. Then, blob analysis was used to filter the small objects which could not be eliminated by morphological filtering and to extract the animal image. Finally, mean color, color variance, rectangularity and area of objective's image are extracted as feature vectors for further animal detection. The results showed that this approach is feasible and effective.