A reference process for automating bee species identification based on wing images and digital image processing

Pollinators play a key role in biodiversity conservation, since they provide vital services to both natural ecosystems and agriculture. In particular, bees are excellent pollinators; therefore, their mapping, classification, and preservation help to promote biodiversity conservation. However, these tasks are difficult and time consuming since there is a lack of classification keys, sampling efforts and trained taxonomists. The development of tools for automating and assisting the identification of bee species represents an important contribution to biodiversity conservation. Several studies have shown that features extracted from patterns of bee wings are good discriminatory elements to differentiate among species, and some have devoted efforts to automate this process. However, the automated identification of bee species is a particularly hard problem, because (i) individuals of a given species may vary hugely in morphology, and (ii) closely related species may be extremely similar to one another. This paper proposes a reference process for bee classification based on wing images to provide a complete understanding of the problem from the experts' point of view, and a foundation to software systems development and integration using Internet services. The results can be extended to other species identification and taxonomic classification, as long as similar criteria are applicable. The reference process may also be helpful for beginners in this research field, as they can use the process and the experiments presented here as a guide to this complex activity.