International Journal of Applied Earth Observation and Geoinformation 18 (2012) 428–435 Contents lists available at SciVerse ScienceDirect International Journal of Applied Earth Observation and Geoinformation jo u rn al hom epage: www.elsevier.com/locate/jag External geo-information in the segmentation of VHR imagery improves the detection of imperviousness in urban neighborhoods Klaartje Verbeeck ∗ , Martin Hermy, Jos Van Orshoven Division of Forest, Nature and Landscape, Department of Earth and Environmental Sciences, KU Leuven, Celestijnenlaan 200E-2411, BE-3001 Leuven, Belgium a r t i c l e i n f o Article history: Received 9 September 2011 Accepted 27 March 2012 Keywords: OBIA QuickBird Image segmentation Classification Large scale reference database Hydrology a b s t r a c t Object-based image analysis (OBIA) has become an established way to detect imperviousness and other land cover classes from very high resolution (VHR) multispectral imagery. Data fusion with LiDAR derived digital surface models (DSM) and large scale vectorial datasets containing building footprints and road boundaries have the potential to significantly improve this method. However, the individual contribution of the large scale vectorial dataset remains unclear. In this paper, we studied the improvement of seg- mentation and classification results when including a vectorial dataset in the OBIA. Two slightly different segmentation methods making use of the vectorial dataset (boundary suggestion method and absolute boundary method) are compared with each other, with a per-pixel classification of the image and an OBIA segmentation without the input of a vectorial dataset. The performance of all four segmentation methods was assessed both for per-pixel image classification and for segmentation accuracy. The classifi- cation accuracy was highest for the segmentation method where the vectorial boundaries were absolute (overall accuracy 82%). However, the boundary suggestion method, where segments were smaller than the reference polygons, had the highest segmentation quality. Although differences between the two methods were clear, the differences with the results of the object-based analysis which did not use the vectorial dataset, were even larger. This indicates that the explicit inclusion of a large scale vectorial dataset is beneficial for the segmentation and classification of imperviousness in an urban environment. © 2012 Elsevier B.V. All rights reserved. 1. Introduction Knowledge of the location, extent and physical properties of impervious surfaces is important for hydrological studies due to their impact on stream water quality and runoff volumes (Arnold and Gibbons, 1996; Booth and Jackson, 1997; Jennings and Jarnagin, 2002). The focus of early research was mainly on the impact of imperviousness on watershed scale, while more recently atten- tion shifted to large scale assessments of smaller areas, especially within built-up environments (Xiao et al., 2007; Mitchell et al., 2008; Perry and Nawaz, 2008). Although individually small and seemingly without effects, the cumulative area of imperviousness in residential gardens can have a large impact on the local water balance (Perry and Nawaz, 2008). Therefore, we are interested in all impervious entities on the parcel both large and small, from large houses to narrow pathways. Object-based image analysis (OBIA) has shown its potential to detect and classify imperviousness in urban environments using ∗ Corresponding author. Tel.: +32 16 329721; fax: +32 16 329760. E-mail addresses: Klaartje.Verbeeck@ees.kuleuven.be (K. Verbeeck), Martin.Hermy@ees.kuleuven.be (M. Hermy), Jos.VanOrshoven@ees.kuleuven.be (J. Van Orshoven). very high resolution imagery (e.g. Mathieu et al., 2007a; Zhou and Troy, 2008; Wurm et al., 2010; Myint et al., 2011). In the object- based approach of image classification, the image is first segmented into spatial objects that are internally relatively homogeneous. These objects are richer in spectral (mean values, minimum and maximum values per band, etc.) and spatial information (shape, distances, context, etc.) compared to the single pixels they consist of. The segmentation step is the first and foremost important step in OBIA, as the accuracy of the classification of the objects is highly dependent on the quality of the segmentation (Neubert et al., 2008). Within the numerous algorithms that exist for image segmentation (Blaschke et al., 2004; Meinel and Neubert, 2004; Blaschke, 2010), the multi-resolution segmentation based on the fractal net evolu- tion approach, implemented in the eCognition software (Definiens, 2010), is a widely accepted segmentation algorithm used for land cover detection in urban environments (e.g. Herold et al., 2002; Van de Voorde et al., 2004; Mathieu et al., 2007a,b; Im et al., 2008; Platt and Rapoza, 2008; Zhou et al., 2008). This segmentation tech- nique has the advantage of being able to produce segments of various sizes within one step as it is based on equal within-object homogeneity rather than size. However, there is no standardized or widely accepted guideline to optimize the homogeneity set- ting (scale parameter) of the segmentation (Carleer et al., 2005; Jacquin et al., 2008; Kim et al., 2008; Myint et al., 2011). As such 0303-2434/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jag.2012.03.015