Rudolf Ressel

A Neural Network-Based Classification for Sea Ice Types on X-Band SAR Images

We examine the performance of an automated sea ice classification algorithm based on TerraSAR-X ScanSAR data. In the first step of our process chain, gray-level co-occurrence matrix(GLCM)-based texture features are extracted from the image. In the second step, these data are fed into an artificial neural network to classify each pixel. Performance of our implementation is examined by utilizing a time series of ScanSAR images in the Western Barents Sea, acquired in spring 2013. The network is trained on the initial image of the time series and then applied to subsequent images. We obtain a reasonable classification accuracy of at least 70% depending on the choice of our ice-type regime, when the incidence angle range of the training data matches that of the classified image. Computational cost of our approach is sufficiently moderate to consider this classification procedure a promising step toward operational, near-realtime ice charting.

Comparing Near Coincident Space Borne C and X Band Fully Polarimetric SAR Data for Arctic Sea Ice Classification

This work compares the polarimetric backscatter behavior of sea ice in spaceborne X-band and C-band Synthetic Aperture Radar (SAR) imagery. Two spatially and temporally coincident pairs of fully polarimetric acquisitions from the TerraSAR-X/TanDEM-X and RADARSAT-2 satellites are investigated. Proposed supervised classification algorithm consists of two steps: The first step comprises a feature extraction, the results of which are ingested into a neural network classifier in the second step. Based on the common coherency and covariance matrix, we extract a number of features and analyze the relevance and redundancy by means of mutual information for the purpose of sea ice classification. Coherency matrix based features which require an eigendecomposition are found to be either of low relevance or redundant to other covariance matrix based features, which makes coherency matrix based features dispensable for the purpose of sea ice classification. Among the most useful features for classification are matrix invariant based features (Geometric Intensity, Scattering Diversity, Surface Scattering Fraction). This analysis reveals analogous results for all four acquisitions, in both X-band and C-band frequencies. The subsequent classification produces similarly promising results for all four acquisitions. In particular, the overlapping image portions exhibit a reasonable congruence of detected ice types.

A Combination of Traditional and Polarimetric Features for Oil Spill Detection Using TerraSAR-X

Synthetic aperture radar (SAR) images are operationally used for the detection of oil spills in the marine environment, as they are independent of sun light and weather-induced phenomena. Exploitation of radar polarimetric features for operational oil spill detection is relatively new and until recently those properties have not been extensively exploited. This paper describes the development of a oil spill detection processing chain using coherent dual-polarimetric (copolarized channels, i.e., HH-VV) TerraSAR-X images. The proposed methodology focuses on offshore platform monitoring and introduces for the first time a combination of traditional and polarimetric features for object-based oil spill detection and look-alike discrimination. A total number of 35 feature parameters were extracted from 225 oil spills and 26 look-alikes and divided into training and validation dataset. Mutual information content among extracted features have been assessed and feature parameters are ranked according to their ability to discriminate between oil spill and look-alike. Extracted features are used for training and validation of a support vector machine-based classifier. Performance estimation was carried out for the proposed methodology on a large dataset with overall classification accuracy of 90% oil spills and 80% for look-alikes. Polarimetric features such as geometric intensity, copolarization power ratio, span proved to be more discriminative than other polarimetric and traditional features.

Investigation into Different Polarimetric Features for Sea Ice Classification Using X-Band Synthetic Aperture Radar

Satellite-borne synthetic aperture radar has proven to be a valuable tool for sea ice monitoring for more than two decades. In this study, we examine the performance of an automated sea ice classification algorithm based on polarimetric TerraSAR-X images. In the first step of our approach, we extract 12 polarimetric features from HH-VV dualpol StripMap images. In a second step, we train an artificial neural network, and then, feed the feature vectors into the trained neural network to classify each pixel into an ice type. The first part of our analysis addresses the predictive value of different subsets of features for our classification process (by means of measuring mutual information). Some polarimetric features such as polarimetric span and geometric intensity are proven to be more useful than eigenvalue decomposition based features. The classification is based on and validated by in situ data acquired during the N-ICE2015 field campaign. The results on a TerraSAR-X dataset indicate a high reliability of a neural network classifier based on polarimetric features. Performance speed and accuracy promise applicability for near real-time operational use.

Automated Iceberg Detection Using High-Resolution X-Band SAR Images

Abstract. In northern latitudes, icebergs frequently cross shipping routes and impair marine traffic. To improve ship routing, we explore the capabilities of an algorithm that detects and charts icebergs from images provided by the German radar satellite TerraSAR-X. TerraSAR-X is in a near-polar orbit, equipped with an active X-Band radar antenna and, thus, allows monitoring the ocean and frozen waters regardless of cloud cover and darkness. The algorithm we apply is based on the iterative censoring constant false alarm rate (IC-CFAR) detector, which has proven its usefulness for terrestrial target detection already. Unlike the standard approach, we not only estimate statistical properties of open water intensities expressed by a probability density function, but also search for recurring patterns (i.e., waves). This allows discriminating icebergs from most false alarms that arise from rough sea and strong winds. Experiments carried out with a series of HH-polarized TerraSAR-X Stripmap images acquired between 2012 and 2015 confirm that, due to consideration of wave pattern during image processing, the false alarm rate is reduced by a factor of 3.

Arctic Sea Ice Characterization Using RISAT-1 Compact-Pol SAR Imagery and Feature Evaluation: A Case Study Over Northeast Greenland

Synthetic Aperture Radar (SAR) polarimetry has become a valuable tool in space-borne SAR-based sea ice analysis. The two major objectives in SAR-based remote sensing of sea ice are, on the one hand, to have a large coverage and, on the other hand, to obtain a radar response that carries as much information as possible in order to characterize sea ice. Single-polarimetric acquisitions of existing sensors offer a wide coverage on the ground, whereas dual polarimetric or even better fully polarimetric data offer a higher information content, which allows for a more reliable automated sea ice analysis at a cost of smaller swath. In order to reconcile the advantages of fully polarimetric acquisitions with the higher ground coverage of acquisitions with fewer polarimetric channels, hybrid/compact polarimetric acquisitions offer an excellent tradeoff between the mentioned objectives. With the advent of the RISAT-1 satellite platform, we are able to explore the potential of compact dual pol acquisitions for sea ice analysis and classification. Our algorithmic approach for an automated sea ice classification consist of two steps. In the first step, we perform a feature extraction followed by a feature evaluation procedure. The resulting feature vectors are then ingested into a trained artificial neural network classifier to arrive at a pixel-wise supervised classification. We present a comprehensive polarimetric feature analysis and classification results on a dataset acquired off the eastern Greenland coast, along with comparisons of results obtained from near-coincident (spatially and temporally) C -band fully polarimetric imagery acquired by RADARSAT-2.

Comparing automated sea ice classification on single-pol and dual-pol TerraSAR-X data

We compare classification of sea ice based on TerraSAR-X (TS-X) images for single-polarization and dual-polarization imaging modes. A texture based implementation for neural network classification on single-polarized ScanSAR data is presented. Likewise we propose an approach for operational generation of dual-polarized Stripmap data (with a different neural network architecture). Polarimetric feature quality in terms of information content is discussed for the latter implementation. Based on these results, neural network classification is applied to image acquired over Svalbard, Baffin Bay, and the Barents Sea. Our successful results justify to increase efforts into exploring further application potential of a software suite which comprises both algorithms. Such a tool may then provide navigational assistance for maritime users in near-real time.

First Tests on Near Real Time Ice Type Classification in Antarctica

In this paper, we explore the capabilities of an algorithm for ice type classification. Our main motivation and exemplary application was the recent incident of the research vessel Akademik Shokalskiy, which was trapped in pack ice for about two weeks. Strong winds had driven ice floes into a bay, forming an area of pack ice, blocking the ships advancement. High-resolution satellite images helped to assess the ice conditions at the location. To extract relevant information automatically from the images, we apply an algorithm that is aimed to generate an ice chart, outlining the different ice type zones such as pack ice, fast ice, open water. The algorithm is based on texture analysis. Textures are selected that allow recognition of different structures in ice. Subsequently, a neural network performs the classification. Since results are output in near real time, the algorithm offers new opportunities for ship routing in ice infested areas.

Neural network based automatic sea ice classification for CL-pol RISAT-1 imagery

SAR Polarimetry has become a valuable tool in spaceborne SAR based sea ice analysis. The two major objectives in SAR based remote sensing of sea ice is on the one hand to have a large coverage of the imaged ground area, and on the other hand to obtain a radar response that carries as much information as possible. Whereas single-polarimetric acquisitions of existing sensors offer a wide coverage on the ground, dual polarimetric, or even better fully polarimetric data offer a higher information content which allows for a more reliable automated sea ice analysis. In order to reconcile the advantages of fully polarimetric acquisitions with the higher ground coverage of acquisitions with fewer polarimetric channels, hybrid polarimetric acquisitions offer a trade-off between the mentioned objectives. With the advent of the RISAT-1 satellite platform, we are able to explore the potential of hybrid dual pol acquisitions for sea ice analysis and classification. Our algorithmic approach for an automated sea ice classification consists of two steps. In the first step, we perform a feature extraction procedure. The resulting feature vectors are then ingested into a trained neural network classifier to arrive at a pixelwise supervised classification. We present first results on a dataset acquired off the eastern Greenland coast.

Offshore pollution monitoring using fully polarimetric X- and C-band synthetic aperture radar: a near-real-time perspective

Use of polarimetric features for oil spill characterization is relatively new and have not been used for operational services until now. In the last decade, a number of semi-automatic and automatic techniques have been proposed in order to differentiate oil spill and look-alike spots based on single pol SAR data, however these techniques suffer from a high miss-classification rate which is undesirable for operational services. In addition to that, small operational spillages from offshore platforms are often ignored as it appears insignificant on traditional ScanSAR (wide) images. In order to mitigate this situation a major focus of research in this area is the development of automated algorithms based on polarimetric images to distinguish oil spills from look-alikes. This paper describes the development of an automated Near Real Time (NRT) oil spill detection processing chain based on quad-pol RADARSAT-2 and quad-pol TerraSAR-X images using polarimetric features (e.g. Lexicographic and Pauli Based features). Number TerraSAR-X images acquired over known offshore platforms with same day ascending and descending configuration along with near coincident RADARSAT-2 acquisition. A total number of 10 polarimetric feature parameters were extracted from different types of oil (e.g. crude oil, emulsion etc) and look-alike spots and divided into training and validation dataset. Extracted features were then used for training and validation of a pixel based Artificial Neural Network (ANN) classifier. Initial performance estimation was carried out for the proposed methodology in order to evaluate its suitability for NRT operational service. Mutual information contents among extracted features were assessed and feature parameters were ranked according to their ability to discriminate between oil spill and look- alike. Polarimetric features such as Scattering diversity, Surface scattering fraction, Entropy and Span proved to be more discriminative than other polarimetric features.