Robust Detection of Water Sensitive Papers

• Tipo: Artigo em Livro de Atas de Conferência Internacional
• Data: 2018

Autores

marcal, ars (Autor) FCUP

Ata de Conferência Internacional

Título: ICIAR

Páginas: 218-226

Indexação

Scopus - 0 Citações

Outras Informações

ID Authenticus: P-00P-1V5

DOI: 10.1007/978-3-319-93000-8_25

Abstract (EN): The automatic analysis of water-sensitive papers (WSP) is of great relevance in agriculture. SprayImageMobile is a software tool developed for mobile devices (iOS) that provides full processing of WSP, from image acquisition to the final reporting. One of the initial processing tasks on SprayImageMobile is the detection (or segmentation) of the WSP on the image acquired by the device. This paper presents the method developed for the detection of the WSP that was implemented in SprayImageMobile. The method is based on the identification of reference points along the WSP margins, and the modeling of a quadrilateral that takes into account possible false positive and negative identifications. The method was tested on a set of 360 images, failing to detect the WSP in only 1 case (detection accuracy of 99.7%). The segmentation accuracy was evaluated using references obtained by a semi-automatic method. The average values obtained for the 359 images tested were: 0.9980 (precision), 0.9940 (recall) and 0.9921 (Hammoude metric). © 2018, Springer International Publishing AG, part of Springer Nature.

Idioma: Inglês

Tipo (Avaliação Docente): Científica