Hélène Sportouche

Extraction and Three-Dimensional Reconstruction of Isolated Buildings in Urban Scenes From High-Resolution Optical and SAR Spaceborne Images

In this paper, we propose a new complete semi-automatic processing chain able to provide, from a couple of high-resolution optical and synthetic aperture radar (SAR) images, a simple 3-D reconstruction of buildings in urban scenes. A sequence of processing, exploring the complementarities of both optical and SAR data, is developed for building reconstruction. The chain is decomposed into the main following steps: First, potential building footprints are extracted from the monoscopic optical image through global detection followed by a boundary refinement. Then, the optical footprints are projected and registered into SAR data to get a fine superposition between optical and SAR homologous ground features. Finally, the last step, based on the optimization of two SAR criteria, is performed to deal with building validation and height retrieval. Each of these steps is methodologically described and applied on scenes of interest on Quickbird and TerraSAR-X images. A qualification of each reconstructed building by a score of confidence is then proposed. Good results of building detection are obtained, and relevant height estimations are retrieved.

Building extraction and 3D reconstruction in urban areas from high-resolution optical and SAR imagery

In this paper, we propose to investigate the joint use of high-resolution optical and SAR data, for building extraction and 3D reconstruction in large urban areas. A sequence of methods providing, in a semi-automatic way, the building detection and reconstruction is presented. Potential building footprints are first extracted on an optical image by a two-phases process (coarse detection and boundaries refinement). The framework of fusion with SAR data is then developed. A first step of registration allows us to get a fine superposition of optical and SAR building features. Then, we show how to take benefit from the introduction of SAR data: The proposed methodology, based on the optimization of criteria refering to building SAR characteristics, allows us to simultaneously deal with the building presence validation and with the height retrieval.

Building detection by fusion of optical and SAR features in metric resolution data

In this paper, we propose to jointly use optical and SAR features issued from satellite images with metric resolution, to deal with the problem of building detection and height retrieval. In a first part, a process — described in previous works — for building boundary extraction, is briefly exposed and illustrated on a Quickbird urban scene. In a second part, the framework of fusion with SAR data is developed. After the steps of feature projection and registration, a new method for building height estimation is proposed. This one is based on a Likelihood criterion optimization and is built on the scheme ”height hypothesis — characteristic areas generation — energy minimization”. Such an approach refers to the adequation between a potential building signature and the real signature, effectively present on the SAR image and defined by characteristic building areas such as layover, shadow, roof and ground / wall echo. This height retrieval process is tested on simulated and real (TerraSAR-X) data.

Building detection and height retrieval in urban areas in the framework of high resolution optical and SAR data fusion

In this paper, we propose a symmetrized version of a semiautomatic processing chain, able to provide the simple 3D reconstruction of buildings in urban scenes, from high-resolution optical and SAR imagery. The new elaborated chain gives an equivalent part to the optical and SAR components, in order to fully exploit complementary information provided by proper building features in both images.

How to combine TerraSAR-X and Cosmo-SkyMed high-resolution images for a better scene understanding?

This paper presents a processing chain to combine a CosmoSkyMed (CSK) image and a TerraSAR-X (TSX) image. After registration and calibration steps, processing at different levels is studied: pixel level for the detection of stable features and joint filtering, primitive level for stability analysis and object level (like roads) for joint interpretation.

Change detection in multitemporal HR SAR images: A hypothesis test-based approach

Starting with a comparison of state of the art criteria and the influence of filtering, this paper presents an analysis tool for multi-temporal SAR images. It is based on the detection of a step change pattern with a generalized maximum likelihood ratio test. Compared to previous works on similar subject, the proposed approach takes into account a filtering step by a non-local approach and a spatially varying equivalent number of looks. Performance analysis is first done on synthetic data and then results on a real data set are analyzed.

A symmetric scheme for building reconstruction from a couple of HR optical and SAR data

In this paper, we propose a new version of a complete processing chain for simple building reconstruction, from a couple of HR optical and SAR satellite images.

Mimick capacity of Generalized Gamma distribution for high resolution SAR image statistical modeling

In this paper we investigate the capacity of the Generalized Gamma distribution to mimick (or imitate) thanks to its three parameters other useful SAR distributions. We first compare it with the Fisher distribution when mimicking a K distribution of reference, thanks to the log-cumulant approach and through a Kullback-Leibler divergence. We then study how the Generalized Gamma distribution can imitate a Log-Normal distribution as asymtotic limit.