analise orientada a objeto

Anais XVI Simpósio Brasileiro de Sensoriamento Remoto - SBSR, Foz do Iguaçu, PR, Brasil, 13 a 18 de abril de 2013, INPE

Dados LiDAR e análise orientada a objeto no monitoramento de manejo florestal
Luciana Spinelli Araujo 1
Michael Keller 1,2
Marcus Vinicio Neves d’Oliveira 3
Daniel Victoria 1
Douglas Morton 4
Celina Maki Takemura 1
Maria O'Healy Hunter 5
1

Embrapa Monitoramento por Satélite
Av. Soldado Passarinho, 303 - 13070-115 - Campinas - SP, Brasil
{spinelli, daniel, celina}@cnpm.embrapa.br
2

USDA Forest Service, International Institute of Tropical Forestry
Jardín Botánico Sur, 1201 Calle Ceiba
San Juan, Puerto Rico, 00926 USA [e-mail] 3

Embrapa Acre
Rodovia BR-364, km 14 - 69900-056 - Rio Branco - AC, Brasil [e-mail] 4

Goddard Space Flight Center
8800 Greenbelt Rd. - Greenbelt - Md., 20771, EUA [e-mail] 5

University of New Hampshire
105 Main Street Durham, NH 03824, EUA [e-mail] Abstract. Although selective logging activity has declined significantly in the past decade, it is still an important land use that affects biological diversity and carbon stocks in the Brazilian Amazon. We present initial results of the application of an object-oriented analysis of airborne small-footprint lidar data as a tool for monitoring forest management in the Antimary State Forest in Acre. For a trial area logged within a few months prior to data acquisition, we processed the lidar returns to calculate the relative density of returns between 0 and 1 m height.
The resulting product highlights areas with bare-ground or minimal ground vegetation allowing us to visualize the network of roads and skid trails beneath the forest canopy. The product was segmented and classified into two categories: (1) roads and skid trails and (2) and all other. The resulting classification was compared to a reference data set developed by visual interpretation and validated by GPS ground control points. We found that
67.0% of the areas