Multispectral aerial images to phenotype yield potential and tree inventory mapping: case studies in dry pea (Pisum sativum) and apple (Malus domestica) nursery

• Ano de defesa: 2017
• Autor(a) principal: Quiros Vargas, Juan Jose
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Fenotipagem; High-throuhput Phenotyping; Índice de vegetação; Inventário; Inventory; Remote sensing; Sensoriamento remoto; Vegetation index
• Link de acesso: http://www.teses.usp.br/teses/disponiveis/11/11152/tde-28022018-180550/

Field data collection involves time and money consuming processes, additionally carrying possible measurement errors. With the technological advance in the last years, low cost remote sensing tools have emerged to facilitate procedures for in-field measurements, being one of the most known techniques the use of multispectral cameras coupled to RPA. These tools are complemented by the implementation of procedures in GIS and image-processing software, from which are developed methodologies leading to extract target values from a certain original set of data. In this work, multispectral images were used in two case studies: (1) for yield estimation in pea plots for breeding research, and (2) for plant counting in an apple nursery planted directly on the soil; both fields are located in Washington State, USA. In the first case, a reliable and replicable methodology for yield estimation was created as a high throughput phenotyping technique; while in the second case an algorithm capable of identifying the number of apple plants with more than 95% accuracy was developed. In both studies, remote sensing is used as an efficient and practical way to improve field operations under the specified conditions of each case.