Pixelwise Time Series Retrieval in Phenological Studies

• Autor(a) principal: Santos, Elisangela Silva
• Data de Publicação: 2019
• Outros Autores: Alberton, Bruna [UNESP], Morellato, Leonor Patricia [UNESP], Da Silva Torres, Ricardo
• Tipo de documento: Artigo de conferência
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1109/IGARSS.2019.8898112; http://hdl.handle.net/11449/232954
• topic: information retrieval; phenology; recurrence plot; time series retrieval

The support of time series similarity searches might be crucial in phenology studies, in which long-term time series analysis based on the identification of similar and different phenological patterns shared by individuals belonging to different species is a widely common task. In this paper, we introduce the use of well-established Information Retrieval (IR) technologies in the search of time series. The solution comprises four main steps: extraction of an image-based time series representation; image content description to encode time series properties and patterns; textual signature extraction based on image content descriptions; and textual signature indexing using off-the-shelf IR approaches. In this paper, we demonstrate both the effectiveness and the efficiency of the proposed solution in time series retrieval problems related to the management of phenological data associated with near-surface vegetation images.