TY - GEN T1 - OBIA and Data Fusion Techniques for Classification and Delineation of Evergreen and Deciduous Tree Canopy AU - Dees, John P. AB - Remote sensing technology has revolutionized the management of land use, urban environments, and natural resources. Massive datasets from satellite imagery, aerial imagery platforms, LiDAR (Light Detection and Ranging), and a variety of other sources are now either publically or commercially available for environmental analysis. Each of these data types offers unique opportunities and challenges for environmental research. Traditional methods of land cover analysis have focused on pixel-level data from a variety of multispectral sensors (pixels store reflectance values of various wavelengths of the electromagnetic spectrum—red, green, blue, infrared, etc.). A more recent thread of research in remote sensing employs Object-Based Image Analysis (OBIA) methods whereby software and user-generated algorithms work in tandem to separate imagery into meaningful objects; OBIA attempts to mimic the object-distinguishing capabilities of human sight. OBIA software such as eCognition Developer has the capability to fuse multiple data types as context into object analysis. This data fusion capability represents a powerful tool for producing high-accuracy classifications. This project proposes to use OBIA and data fusion to delineate evergreen and deciduous tree canopy within Hall County, Georgia. Two sets of high-resolution (seasonal leaf-on and leaf-off aerial) imagery will be used along with elevation datasets generated from LiDAR data (normalized Digital Surface Model). An algorithm (eCognition ruleset) will be written which utilizes the specific “fingerprint” (height, spectral characteristics, and change detection in seasonal imagery) of each tree canopy type to correctly classify land cover within the imagery. DA - 2015-4-1 PY - 2024 PB - unav N1 -

Acknowledgements:

Dr. J.B. Sharma

ER -