The 2013 European Space Agency Living Planet Symposium Edinburgh, United Kingdom from 9 to 13 September 2013 (http://www.livingplanet2013.org/index.asp) A SEMI AUTOMATED OBJECT BASED IMAGE ANALYSIS APPROACH FOR LANDSLIDE DELINEATION Bakhtiar Feizizadeh (1) , Thomas Blaschke (2) (1-2) Centre of Remote and GIS, University of Tabriz, Iran, Email: Feizizadeh@tabrizu.ac.ir (2) Department of Geoinformatics, University of Salzburg, Austria, Email: Thomas.Blaschke@sbg.ac.at ABSTRACT The main objective of this research is to examine a semi-automated Object-based image analysis (OBIA) process for landslides bodies’ delineation in the Urmia lake basin, Northern Iran. We use IRS and SPOT satellite images in combination a DEM, slope, flow direction and curvature maps. A sequence of image segmentation, feature selection, object classification and error balancing was developed and tested on a variety of sample datasets (current landslide inventory) of the study area. In order to compute the accuracy an error matrix and overall accuracy were calculated. An overall accuracy was about 93.07 % which reveals the high reliability of obtained landslide inventory map. The results confirm the potential of OBIA to derive detailed and accurate information for landslide delineating from high resolution satellite images at multiple scales and the results highlight important landslide monitoring and management implications. We conclude that there is a potentiality to improve the accuracy either by considering further parameters during rulebased classification or using satellite images with a higher spatial resolution. 1. INTRODCTION Based on the recently progress in earth observation and in particularly remote sensing satellite imagery a Geoinformation plays a key role for detecting, analyzing and monitoring landslides to assist hazard and risk analysis (Hölbling et al., 2012). Satellite remote sensing technology has proven to be the best tool for generating such landslide inventories, particularly with the availability of high-resolution images (Chen et al., 2007; Martha, 2011). In general, remote sensing techniques can be used in landslide studies in detecting, monitoring, and classifying landslides such as aerial photography interpretations, stereoscopic image analysis, and interferometry studies (Aksoy and Ercanoglu, 2012). Today, the wide range of available earth observation data implies the need for accurate and fast methods for detecting, analyzing and monitoring landslides and to facilitate the generation of landslide inventory maps and databases to assist hazard and risk analysis (Hölbling et al., 2012). Object-based image analysis (OBIA) has gained prominence in the field of remote sensing during the last decade, being credited with the potential of overcoming weaknesses associated with the per pixel analysis, as for instance neglecting geometric and contextual information (Blaschke, 2010; Draguţ and Eisank, 2012). In the context of image processing methods to detect and delineate of landslides, OBIA have a high potential to monitor the evolution of landslide-prone areas over time, as spectral, spatial, contextual as well as morphological parameters can be considered (Hölbling and Füreder, 2001; Hölbling et al., 2012). OBIA has already proven its ability for successful automatic classification of landforms and landslides (Martha et al., 2010). OBIA considered as a platform for integration of spectral and spatial data (e.g. elevation and thematic) (Dragut and Blaschke, 2006; van Asselen and Seijmonsbergen, 2006; Martha et al., 2010). OBIA, and to some extent GEOBIA (geographic object-based image analysis) (Blaschke, 2010), is a knowledge-driven method, whereby spectral, morphometric, and contextual diagnostic features of an object can be integrated based on expert knowledge (Barlow et al., 2003; Aksoy and Ercanoglu, 2012). It allows the user to apply locally different strategies for analyses. Incorporating both spectral information (e.g. DN values, tone and color) and spatial arrangements (e.g. environmental information, size, shape, texture, pattern, and association with neighboring objects) comes closer to the way humans interpret information visually from aerial photos (Laliberte et al., 2004; Aksoy and Ercanoglu, 2012). Since OBIA take the integration of spectral and spatial information in to account of classification, consequently it has a high potential to detect landslides automatically in a better way than the pixel-based methods, by incorporating a multitude of landslide diagnostic features (Martha et al., 2010). Object based classification is a knowledge driven method, whereby spectral, morphometric and contextual landslide diagnostic features can be integrated based on expert knowledge to accurately detect landslides (Barlow et al., 2003, 2006; Martha et al. 2010). In despite of OBIA’s high potentialities to integrate remote sensing and GIS dataset for landslide studies, OBIA based landslides detection has been