Sascha Klonus

Multi-sensor image fusion for pansharpening in remote sensing

The main objective of this article is quality assessment of pansharpening fusion methods. Pansharpening is a fusion technique to combine a panchromatic image of high spatial resolution with multispectral image data of lower spatial resolution to obtain a high-resolution multispectral image. During this process, the significant spectral characteristics of the multispectral data should be preserved. For images acquired at the same time by the same sensor, most algorithms for pansharpening provide very good results, i.e. they retain the high spatial resolution of the panchromatic image and the spectral information from the multispectral image (single-sensor, single-date fusion). For multi-date, multi-sensor fusion, however, these techniques can still create spatially enhanced data sets, but usually at the expense of the spectral consistency. In this study, eight different methods are compared for image fusion to show their ability to fuse multitemporal and multi-sensor image data. A series of eight multitemporal multispectral remote sensing images is fused with a panchromatic Ikonos image and a TerraSAR-X radar image as a panchromatic substitute. The fused images are visually and quantitatively analysed for spectral characteristics preservation and spatial improvement. It can not only be proven that the Ehlers fusion is superior to all other tested algorithms, it is also the only method that guarantees excellent colour preservation for all dates and sensors used in this study.

Image Fusion Using the Ehlers Spectral Characteristics Preservation Algorithm

Image fusion is a technique that is used to combine the spatial structure of a high-resolution panchromatic image with the spectral information of a low- resolution multispectral image to produce a high-resolution multispectral image. Currently, image fusion techniques via color or statistical transforms such as the Intensity-Hue-Saturation (IHS) and principal component (PC) methods are still widely used. These methods create multispectral images of higher spatial resolution but usually at the cost of color distortions in the fused images. This is especially true if the wavelength range of the panchromatic image does not correspond to that of the employed multispectral bands or for multitemporal/multisensoral fusion. To overcome the color distortion problem, a number of new fusion methods have been developed over the last years. One of these is the Ehlers fusion algorithm, which is based on an IHS transform coupled with adaptive filtering in the Fourier domain. This method preserves the spectral characteristics of the lower spatial resolution multispectral images for single-sensor, multi-sensor, and multi-temporal fusion. A comparison between this method and three sophisticated new fusion techniques that are available in commercial image processing software is presented in this paper using multitemporal multi-sensor fusion with SPOT multispectral and Ikonos panchromatic datasets as well as single-sensor single-date multispectral and panchromatic Quick-bird data. The fused images are compared visually and with statistical methods that are objective, reproducible, and quantitative. It can be shown that the sophisticated methods such as Gram Schmidt fusion, CN spectral sharpening, and the modified IHS provide good results in color preservation for single sensor fusion. For multi-temporal multi-sensor fusion, however, these methods produce significant changes in spectral characteristics for the fused datasets. This is not the case for the Ehlers fusion algorithm, which shows no recognizable color distortion even for multi-temporal and multi-sensor datasets.

Combined Edge Segment Texture Analysis for the Detection of Damaged Buildings in Crisis Areas

This paper describes the results of a new combined method that consists of a cooperative approach of several different algorithms for automated change detection. These methods are based on isotropic frequency filtering, spectral and texture analysis, and segmentation. For the frequency analysis, different band pass filters are applied to identify the relevant frequency information for change detection. After transforming the multitemporal images using a fast Fourier transform and applying the most suitable band pass filter to extract changed structures, we apply an edge detection algorithm in the spatial domain. For the texture analysis, we calculate the parameters energy and homogeneity for the multitemporal datasets. Then a principal component analysis is applied to the new multispectral texture images and subtracted to get the texture change information. This method can be combined with spectral information and prior segmentation of the image data as well as with morphological operations for a final binary change result. A rule-based combination of the change algorithms is applied to calculate the probability of change for a particular location. This Combined Edge Segment Texture (CEST) method was tested with high-resolution remote-sensing images of the crisis area in Darfur (Sudan). Our results were compared with several standard algorithms for automated change detection, such as image difference, image ratio, principal component analysis, multivariate alteration detection (MAD) and post classification change detection. CEST showed superior accuracy compared to standard methods.

Potential of Digital Sensors for Land Cover and Tree Species Classifications - a Case Study in the Framework of the DGPF- Project

Summary: The study is intended as a contribution to assessing the value of digital image data for semi-automatic analysis for classifying land cover and tree species and was carried out in the framework of the DGPF-project. Sensor specific strengths of ADS40-2 nd , Quattro DigiCAM, DMC, JAS-150, Ultracam-X, and RMK-Top15 cameras and weakness for classification purposes are presented and shortly discussed. The first approach is based on a maximum likelihood method in combination with a decision tree and produces 13 land cover classes. The second approach is based on logistic regression models and produces eight tree species classes. The classified images were visually assessed and quantitatively analyzed. The accuracy assessment reveals that in both approaches similar classification results are obtained by all sensors with overall Kappa coefficients between 0.6 and 0.9. However, a real sensor comparison was not possible since the image data was acquired at different dates. Thus, some variations in classification results are due to phenological differences and different illumination and atmospheric conditions. It is planned for the future that the classifications of the first approach will be adjusted to the characteristics of each sensor. In the second approach, further work is needed to improve distinguishing non-dominant, small and covered deciduous tree species. Zusammenfassung: Potenzial digitaler Sensoren zur Klassifizierung der Landbedeckung und Baumarten - eine Fallstudie im Rahmen des DGPF-Projektes. Anhand der Bilddaten aus den Kamerasystemen ADS40-2 nd , Quattro DigiCAM, DMC, JAS-150, Ultracam-X, und RMK-Top15 wurden zwei Klassifikationsverfahren (Maximum Likelihood und logistische Regression) getestet. Dabei wurden sensor-spezifische Eigenschaften erlautert, sowie die Starken und Schwachen der einzelnen Systeme aufgezeigt. Die Resultate wurden visuell und quantitativ bewertet. Direkte Sensorvergleiche erwiesen sich dabei als schwierig, da zum Aufnahmezeitpunkt der einzelnen Bilddaten sowohl eine unterschiedliche Vegetationsentwicklung wie auch Unterschiede in den Beleuchtungs- und atmospharischen Verhaltnissen vorherrschten. Quantitative Analysen zeigen, dass sich mit jedem Kamerasysteme sehr ahnlich gute Resultate erzielen liessen. Das erste Verfahren zeigt fur 13 Landnutzungsklassen Kappa Koeffizienten von gut 0.6 bei allen verwendeten Systemen. Allerdings unterscheidet sich die Genauigkeit der einzelnen spezifischen Klassen wie Mais oder Kartoffeln fur die unterschiedlichen Kameras. Hierzu soll in weiteren Analysen das Klassifikationsverfahren an die jeweiligen Kameras angepasst werden. Fur das zweite Verfahren liegt der Kappa Koeffizient fur 8 Baumarten zwischen 0.7 und 0.9. Bei diesem Verfahren soll in zukunftigen Analysen die Genauigkeit der Erkennung von nicht dominanten, kleinen und verdeckten Baumarten erhoht werden.

Colour and texture based change detection for urban disaster analysis

A rapid visualisation of change in urban crisis areas is an important condition for planning and coordination of help. For automated change detection, a large number of algorithms has been proposed and developed. This paper describes the results of a colour and texture based change detection approach that was applied to satellite and aircraft images of the earthquake region in Haiti. In our integrated methodology, we calculate firstly new colour texture images which are based on the feature ‘energy’. This is performed for every channel in the visible colour spectrum at two different times after a radiometric harmonisation. Combined with a principal component analysis for each texture image and a subsequent histogram optimisation, we subtract the texture elements to visualise the occurred change. As result, it is not only possible to automatically delineate areas of change but also to distinguish between different types of change.

Synergetic use of TerraSAR-X and Radarsat-2 time series data for identification and characterization of grassland types - a case study in Southern Bavaria, Germany

In the context of global change, alteration of landscapes and loss of biodiversity, the monitoring of habitats, vegetation types and their changes have become extraordinary important. In this paper, first results from a study that analyses the differentiability of NATURA2000 habitats and HNV grassland with imaging radar data are presented. Therefore, Kennaugh elements derived from TerraSAR-X and Radarsat-2 dual pol (VV/VH) time series data are used, both separately and in combination, to model the distribution of these classes with the Maximum-Entropy principle. The preliminary results show that the multi-frequency approach enables - compared to single frequency analyses - a finer differentiation between scatterers in the size of 3-6 cm (e.g. 7120, 7230 and HNV grassland).

Quality Assessment for Multitemporal and Multi Sensor Image Fusion

Generally, image fusion methods are classified into three levels: pixel level (iconic), feature level (symbolic) and knowledge or decision level. In this paper we focus on iconic techniques for image fusion. There exist a number of established fusion techniques that can be used to merge high spatial resolution panchromatic and lower spatial resolution multispectral images that are simultaneously recorded by one sensor. This is done to create high resolution multispectral image datasets (pansharpening). In most cases, these techniques provide very good results, i.e. they retain the high spatial resolution of the panchromatic image and the spectral information from the multispectral image. These techniques, when applied to multitemporal and/or multisensoral image data, still create spatially enhanced datasets but usually at the expense of the spectral consistency. In this study, a series of nine multitemporal multispectral remote sensing images (seven SPOT scenes and one FORMOSAT scene) is fused with one panchromatic Ikonos image. A number of techniques are employed to analyze the quality of the fusion process. The images are visually and quantitatively evaluated for spectral characteristics preservation and for spatial resolution improvement. Overall, the Ehlers fusion which was developed for spectral characteristics preservation for multi-date and multi-sensor fusion showed the best results. It could not only be proven that the Ehlers fusion is superior to all other tested algorithms but also the only one that guarantees an excellent color preservation for all dates and sensors.

Detektion von zerstörten Gebäuden in Krisengebieten aus pan-chromatischen Daten

Das Ziel dieses Artikels ist die Analyse von Veranderungen in Gebieten, in denen sich Katastrophen mit plotzlichen Anderungen an Gebauden und der Infrastruktur ereignet haben. Standardverfahren der Veranderungsanalyse fuhren zu keinem zufriedenstellenden Ergebnis, daher wurde ein neues Verfahren entwickelt. Die in diesem Artikel dargestellte Methode erlaubt eine schnelle Detektion und Visualisierung von Veranderungen in Krisen- und Katastrophengebieten. Dies ist eine wichtige Voraussetzung fur die Planung und Koordination von Hilfskrafteinsatzen. Die vorgeschlagene Methode basiert auf Frequenzanalysen, Segmentierung und Texturmerkmalen. Sie kombiniert die unterschiedlichen Ansatze in einem Verfahren mittels eines Entscheidungsbaumes. Im Vergleich mit funf Standardverfahren zeigte dieser neue Ansatz die besten Resultate.

Interpretability of TerraSAR-X fused data

Pansharpening is an image fusion technique that combines the spatial structure of a high resolution panchromatic image with the spectral information of a lower resolution multispectral image to produce a high resolution multispectral image. Image data of the new German RADAR satellite TerraSAR-X were used to sharpen optical multispectral data. To assess the advantages and limitations of fusion, the interpretability of terrain features at different image resolutions was determined. We concluded that a resolution ratio of 1:10 (TerraSAR:Multispectral) is optimal to benefit from the synergism of a SAR/multispectral fusion. Q index and an object-based classification were used to assess fusion quality and to compare their efficacy to determine the best fusion algorithms. Both approaches are appropriate methods to asses quality but only on judging some aspects of the fusion products. We conclude that a more comprehensive fusion quality assessment method still needs to be developed.