Agricultural crops mapping from medium spatial resolution data using object-based analysis

• Grant number: 09/02037-3
• Support type: Regular Research Grants
• Duration: June 01, 2009 - May 31, 2010
• Field of knowledge: Agronomical Sciences - Agronomy
• Principal researcher: Antonio Roberto Formaggio
• Grantee: Antonio Roberto Formaggio
• Home Institution: Instituto Nacional de Pesquisas Espaciais (INPE). Ministério da Ciência, Tecnologia e Inovações (Brasil). São José dos Campos , SP, Brazil

Abstract

Agriculture exerts an important role in the Brazilian socio-economic scenario. Accordingly, government measures have to be supported by tools that provide consistent and reliable information on the consequences of their adoption. Remote Sensing is one of the tools that provide a rich database for different applications in the agricultural field. However, the conventional systems of satellite imagery classification present shortcomings in the process of automatically extracting information from orbital data. In the case of agriculture, there are several problems caused by spectral variability in the images due to the different conditions and types of crops. This hinders the performance of conventional classification algorithms (pixel to pixel), as they proceed their algorithmic processing guided solely by the statistical variables of the data. An alternative that has emerged are the Knowledge Based Systems (KBS), which have a great potential owing to the strategies they employ for storage and replication of human knowledge. Object-Based Analysis fits within the KBS context, since this approach allows the simulation of visual interpretation through modeling of knowledge and, in order to do that, uses mainly topological (neighborhood, context) and geometric (shape and size) information. This paper proposes the investigation of the potential applications of Object-Based Analysis in agricultural mapping by comparing their performance with that of conventional systems. In order to do that, TM and ETM+/Landsat images acquired in a study area located in the northwest of São Paulo State will be employed. That area represents adequately the conditions of agriculture in São Paulo and much of the southeastern and southern regions in Brazil. As a significant result, it is expected that the completion of the proposed research will demonstrate that the AOO exhibits a higher classification performance than conventional algorithms. (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

VIEIRA, MATHEUS ALVES; FORMAGGIO, ANTONIO ROBERTO; RENNO, CAMILO DALELES; ATZBERGER, CLEMENT; AGUIAR, DANIEL ALVES; MELLO, MARCIO PUPIN. Object Based Image Analysis and Data Mining applied to a remotely sensed Landsat time-series to map sugarcane over large areas. REMOTE SENSING OF ENVIRONMENT, v. 123, p. 553-562, AUG 2012. (09/02037-3)