A method for photo-identification of turtles of the Phrynops williamsi species
Resumo
This paper presents a method for photo-identification of Phrynops williamsi turtles. This identification is performed using shape descriptors which computationally represent the horseshoe-shaped and circular-shaped bands that can be found on the ventral surface of the turtle's head. The input image is converted to grayscale, binarized, filtered with morphologic operations, segmented based on the contours of the object and the components are selected based on their geometric characteristics. With the extraction of these characteristics, the method calculates de Fourier Descriptors and create a unique identifier used to identify the turtle from the input image. Results show that the presented method has reached a success rate of 85.71% in intra-class comparison and 85.17% in inter-class comparison.
Referências
Balestra, Rafael Antônio. Roteiro para Inventários e Monitoramentos de Quelônios Continentais, Biodiversidade Bras. 6 (2016) 114–152.
Beugeling, Trevor; Branzan-Albu, Alexandra Computer Vision-Based Identification of Individual Turtles Using Characteristic Patterns of Their Plastrons, in: 2014 Can. Conf. Comput. Robot Vis., IEEE, 2014: pp. 203–210.
El-ghazal, Akrem; Basir, Otman; Belkasim, S. Farthest point distance: A new shape signature for Fourier descriptors, Signal Process. Image Commun. 24 (2009) 572–586.
McDiarmid, Roy et al. Reptile biodiversity: Standard methods for inventory and monitoring, University of California Press, 2012.
Nixon, Mark; Aguado, Alberto. Elsevier Science Publishers., Feature extraction and image processing, Newnes, 2002.
Pic4Turtle, (2016). www.pic4turtle.com.
Rhodin, Anders G. J; Russel, Anthony. Description of Phrynops williamsi, a new species of chelid turtle of the South American Phrynops geoffroanus complex, in: Adv. Herpetol. Evol. Biol. Essays Honor Ernest E. Williams, Cambridge, Mass. :Museum of Comparative Zoology, 1983: pp. 58–73.
Spier, Edson Fernando et al. Registro de Phrynops williamsi (Rhodin & Mittermeier, 1983) no rio do Peixe, centro-oeste de Santa Catarina, Brasil, Brazilian J. Biosci. 12 (2014) 56–57.
Suzuki, Satoshi, Topological structural analysis of digitized binary images by border following, Comput. Vision, Graph. Image Process. 30 (1985) 32–46.
Vogt, Richard Carl. Avaliação do Risco de Extinção de Phrynops williamsi Rhodin & Mittermeier, 1983 no Brasil, (2015).
