Individual identification of animals is of paramount importance to analyze population size, dispersal, habitat preferences or behaviour. Especially for sensitive, threatened species, it is advisable to develop non-invasive recognition methods avoiding direct handling and tagging of the study subjects to be applied to procedures such as marking-recapture. Here we present an application of the I3S software for the individual recognition of the Rosalia longicorn Rosalia alpina based on the contour digitization of the spots present on the beetle's elytra. Classification performances to individual level tested on an overall sample of 290 images (one per subject) were 94.8 (both elytra), 94.5 (right elytron) and 95.2 % (left elytron). Since I3S leaves the final decision to the operator, such high classification performances may be refined further in the final step leading to a fully reliable identification. We found that the identification performance was statistically supported and that the influence of two main error sources (contour tracing and angle under which the images were taken) was negligible. Our approach minimizes the subjectivity of a qualitative manual comparison of images and greatly reduces the time taken to visually retrieve the image of an individual especially for large photo libraries. It may be successfully used in surveys covering large areas and involving many untrained operators such as volunteers or park rangers. We propose that I3S can be applied to other insect species presenting characteristic spot patterns. To our best knowledge, this is the first study using computer-aided identification of a terrestrial arthropod. © 2013 Springer Science+Business Media Dordrecht.
Spotting the right spot: Computer-aided individual identification of the threatened cerambycid beetle Rosalia alpina
Bosso L.Conceptualization
;
2013
Abstract
Individual identification of animals is of paramount importance to analyze population size, dispersal, habitat preferences or behaviour. Especially for sensitive, threatened species, it is advisable to develop non-invasive recognition methods avoiding direct handling and tagging of the study subjects to be applied to procedures such as marking-recapture. Here we present an application of the I3S software for the individual recognition of the Rosalia longicorn Rosalia alpina based on the contour digitization of the spots present on the beetle's elytra. Classification performances to individual level tested on an overall sample of 290 images (one per subject) were 94.8 (both elytra), 94.5 (right elytron) and 95.2 % (left elytron). Since I3S leaves the final decision to the operator, such high classification performances may be refined further in the final step leading to a fully reliable identification. We found that the identification performance was statistically supported and that the influence of two main error sources (contour tracing and angle under which the images were taken) was negligible. Our approach minimizes the subjectivity of a qualitative manual comparison of images and greatly reduces the time taken to visually retrieve the image of an individual especially for large photo libraries. It may be successfully used in surveys covering large areas and involving many untrained operators such as volunteers or park rangers. We propose that I3S can be applied to other insect species presenting characteristic spot patterns. To our best knowledge, this is the first study using computer-aided identification of a terrestrial arthropod. © 2013 Springer Science+Business Media Dordrecht.File | Dimensione | Formato | |
---|---|---|---|
s10841-013-9561-0.pdf
solo utenti autorizzati
Descrizione: Spotting the right spot: Computer-aided individual identification of the threatened cerambycid beetle Rosalia alpina
Licenza:
NON PUBBLICO - Accesso privato/ristretto
Dimensione
499.92 kB
Formato
Adobe PDF
|
499.92 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.