Comparison of Geo-Object Based and Pixel-Based Change Detection of Riparian Environments using High Spatial Resolution Multi-Spectral Imagery

This is the authors' abstract. We don't add key points for this paper.

The objectives of this research were to (a) develop a geo-object-based classification system for accurately mapping riparian land-cover classes for two QuickBird images, and (b) compare change mops derived from geo-object-based and per-pixel inputs used in three change detection techniques. The change detection techniques included post-classification comparison, image differencing and the tasseled cap transformation. Two QuickBird images, atmospherically corrected to at-surface reflectance, were captured in May and August 2007 for a savanna woodlands area along Mimoso Creek in Central Queensland, Australia. Concurrent in-situ land-cover identification and lidar data were used for calibration and validation. The geo-object-based classification results showed that the use of class-related features and membership functions could be standardized for classiffying the two QuickBird images. The geo-object-based inputs provided more accurate change detection results than those derived from the pixel-based inputs, as the geo-object-based approach reduced mis-registration and shadowing effects and allowed inclusion of context relationships.