Date of Award
2010
Degree Type
Thesis
Degree Name
Master of Science
Program
Geography
Supervisor
Dr. Jinfei Wang
Abstract
Tree canopy cover is a fundamental measure of the urban forest, which benefits a city socially and environmentally. In this thesis, methods are proposed to map urban tree cover. Chapter 2 presents an object-based tree cover extraction method using some new techniques for commonly available high-spatial-resolution colour-infrared imagery. The overall accuracy achieved for the 23 645 ha urban growth area of London, Ontario was 89.73%. This accuracy can be improved further by integrating LiDAR surface information. However, tall objects appear displaced in traditional orthoimages, causing misclassification. Chapter 3 presents a new method for correcting horizontal relief displacement of tall objects in orthorectified imagery without requiring the original aerial images and flight parameters. An object-based tree cover extraction method was developed to test the effectiveness of this correction. The overall accuracy for a 1600 ha sub-scene was improved significantly: from 94.66% (uncorrected) to 96.07% (empirically corrected) and 96.98% (geometrically corrected).
Recommended Citation
Lehrbass, Brad, "OBJECT-BASED URBAN TREE COVER EXTRACTION FROM HIGH SPATIAL RESOLUTION OPTICAL AND LIDAR IMAGERY: TECHNIQUES AND DATA INTEGRATION" (2010). Digitized Theses. 4509.
https://ir.lib.uwo.ca/digitizedtheses/4509