Building Height Retrieval in Urban Areas in the Framework of High Resolution Optical and Sar Data Fusion

During the last years, new approaches, exploring the high detail level characterizing high-resolution optical and SAR images provided by current spaceborne sensors, have been proposed for object detection and reconstruction in urban areas. Especially, challenges are born in the fields of building extraction and building height estimation for 3D reconstruction of urban scenes. Some technics such as stereoscopy [1], radargrammetry [2] [3] and interferommetry [4] [5] or simulation-based methods [6], have been applied to deal with this problematic and have given satisfying results of building extraction and height estimation. Nevertheless, some difficulties can be still pointed out: optical couple availability, need of auxiliary input data, incomplete or noisy results from SAR images... In this context, news methods have been recently developed in the frame of optical SAR data fusion, that focus, most of the time, independently on different individual aspects or subparts of the 3D building reconstruction, such as the building boundary detection [7] [8] [9] and the height estimation from one single SAR image [10] or from a couple of interferometric SAR images [11]. We propose, in this paper, a complete and semi-automatic processing chain, able to provide the simple 3D reconstruction of buildings in large urban scenes. The process takes only as input one HR optical image and one HR SAR image of the same area and tends to fully exploit complementary information provided by proper building features in both data. A simple parallelepipedic building model is used for the reconstruction. The chain is decomposed into the two main steps: In a first step, a new scheme for building footprint extraction based on the combination of optical and SAR features is developed and allows us to obtain a robust building detection, ensured by an optical SAR crossed validation. In a second step, a process for building height estimation from SAR data, developed in our previous works [12], is briefly reminded and a new approach for building presence validation, taking into account confidence scores computed during both steps, is proposed. Finally, the 3D reconstruction of a studied urban scene of interest is achieved by using information obtained throughout the chain about building location, height estimation and presence validation, and is illustrated on a couple of Quickbird and TerraSAR-X images in the area of Marseille (France).