U. Benz

A comparison of several algorithms for SAR raw data compression

Proposes new algorithms for synthetic aperture radar (SAR) raw data compression and compares the resulting image quality with the quality achieved by commonly used methods. The compression is carried out in time and frequency domain, with statistic, crisp, and fuzzy methods. The algorithms in the time domain lead to high resolution and a good signal-to-noise ratio, but they do not optimize the performance of the compression according to the frequency envelope of the signal power in both range and azimuth directions. The hardware requirements for the compression methods in the frequency domain are significant, but a higher performance is obtained. Even with a data rate of 3 bits/sample, a satisfactory phase accuracy is achieved which is an essential parameter for polarimetric and interferometric applications. Preliminary analysis concerning the suitability of the proposed algorithms for different SAR applications shows that the compression ratio should be adaptively selected according to the specific application. >

Supervised fuzzy analysis of single- and multichannel SAR data

The paper proposes a new learning fuzzy classification for single and multichannel synthetic aperture radar (SAR) data. It consists of the fusion of a supervised learning fuzzy distribution estimator and an unsupervised learning fuzzy vector quantizer. The adaptive algorithm accommodates varying requirements and delivers classification results in near real time. In addition to the classification, the user gets the reliability of the classification. This knowledge can be used to fuse several sensor channels efficiently. Automatically, a rule base is developed to deliver the required information with the highest possible reliability. In the authors example, the channels of a full polarimetric SAR are used. However, the algorithm can be extended also to optic and infrared channels. The proposed fuzzy classification system forms one module of an adaptive remote-sensing system. A conceptual design of this system is given. System control relies on an expert knowledge base and allows automatic configuration of the system to the considered remote-sensing application. This will lead to an increased usefulness of remotely sensed data.

Object based analysis of polarimetric SAR data in alpha-entropy-anisotropy decomposition using fuzzy classification by eCognition

Polarimetric SAR data possess a high potential for classification of the Earth surface. Various publications demonstrate detailed analysis of soil and vegetation properties and characteristics of man made structures on selected examples. To ensure wider application of these developments, integration in commercial systems should be studied. In a first approach, the object based image analysis eCognition is employed on alpha, entropy and anisotropy and the span of fully polarimetric L-band SAR data of the German airborne sensor, E-SAR. We show that by using eCognition land cover classes can be conveniently assigned to the scattering classes and ambiguities can be resolved by geometric and context object features.

Efficient SAR raw data compression in frequency domain

SAR raw data compression is necessary to reduce the huge amount of data for downlink and the required memory on board. In view of interferometric and polarimetric applications for SAR data it becomes more and more important to pay attention to phase errors caused by data compression. Here, a detailed comparison of block adaptive quantization in time domain (BAQ) and in frequency domain (FFT-BAQ) is given. Inclusion of raw data compression in the processing chain allows an efficient use of the FFT-BAQ and makes implementation for on-board data compression feasible. The FFT-BAQ outperforms the BAQ in terms of signal-to-quantization noise ratio and phase error and allows a direct decimation of the oversampled data equivalent to FIR-filtering in time domain. Impacts on interferometric phase and coherency are also given.

Wavelet based approaches for efficient compression of complex SAR image data

New wavelet based approaches for efficient data compression of complex SAR images with high reconstruction quality are presented. These approaches utilize either a polar format representation to compress magnitude and phase information of the complex SAR images, separately by different compression schemes, or use a Fourier transform scheme to convert the complex image data format to a real data format before applying the wavelet decomposition technique. In both concepts the phase information of the compressed images is preserved in great extent. Additionally, a concept based on vector quantization is used, taking advantage of the correlation between polarization channels for data compression in multi-polarization channel modes. Therefore, these approaches are suitable for data compression in both interferometric and polarimetric applications. Finally, the quality of the reconstructed images are compared in terms of compression ratio and proved by image quality parameters.

A comparison of several algorithms for on-board SAR raw data reduction

This paper shows a comparison of different SAR (synthetic aperture radar) raw data reduction algorithms as applied to E-SAR data (experimental airborne SAR) and spaceborne ERS1 data. The Block Adaptive Quantizer (BAQ) and a Fuzzy Block Adaptive Quantizer (FBAQ) were selected and analysed. In addition, different algorithms based on a BAQ, the Discrete Cosine Block Adaptive Quantizer (DCT-BAQ), the Walsh-Hadamard Block Adaptive Quantizer (WHT-BAQ) and the Block Adaptive Vector Quantizer (BAVQ) mere examined. Signal-to-distortion noise ratios (SDNR) of 11.69 dB (BAQ), 8.00 dB (FBAQ), 2.09 dB (DCT-BAQ and) and 11.94 dB (BAVQ) for E-SAR data and 8.77 dB (BAQ), 5.17 dB (DCT-BAQ) and 9.56 dB (BAVQ) for ERS1 data for a data rate of 2 bits/sample were achieved with a reduction factor of about 3 for the E-SAR data and 2.5 for ERS1 data.<<ETX>>

Analysis of single and multi-channel SAR data using fuzzy logic

Two main problems reduce the acceptance of SAR sensors: The huge data rate impairs real-time distribution and the interpretation of SAR images requires special SAR knowledge by the user. While scientific applications will still need the whole amount of data, many users will only consider SAR data for their tasks, if they can access pre-classified data easily and in (near) real-time. Therefore, an application dependent SAR classification has to be applied on the data to extract the required information. Classification will not only simplify data interpretation but it will also lead to a significant data compression. This paper proposes a fuzzy system to build such an adaptive remote sensing classification module. Fuzzy logic allows simple algorithms, gives the system a high tolerance to parameter variations and adaptivity can easily be implemented. One SAR channel classification is described. It allows a user to define interactively the data classes of interest and thus realizes a flexible analysis of SAR data. The system adapts itself to the various user requirements. In many cases one data source does not deliver enough information to perform the required classification. Data fusion of several sensors or various sensor channels has to be taken into account. A new approach is presented for the fusion of the classification results of three SAR polarization channels. The algorithm uses the FLVQs output of each channel and consists mainly of a fuzzy rule base. This rule base is adaptive to the data sets and to the user requirements.

Image content dependent compression of polarimetric SAR data

This paper presents a novel technique for flexible SAR image compression in wavelet domain. Two approaches are considered: a) Compression leading to a scene independent, very homogenous signal-to-distortion noise ratio and constant phase error. Reconstruction quality varies-dependent from the required compression ratio-from good visible quality to nearly lossless reconstruction. b) Compression leading to high reconstruction quality for regions of interest and high compression ratio on background regions.

Comparison of several algorithms for on-board SAR raw data reduction

This paper gives a comparison of different SAR (Synthetic Aperture Radar) raw data reduction algorithms as applied to E-SAR data (Experimental airborne SAR) and spaceborne ERS1 data. The Block Adaptive Quantizer (BAQ) and a Fuzzy Block Adaptive Quantizer (FBAQ) were selected and analyzed. In addition, different algorithms based on a BAQ, the Fast Fourier Block Adaptive Quantizer (FFT-BAQ) and the Block Adaptive Vector Quantizer (BAVQ) were examined. Signal-to-distortion noise ratios (SDNR) of 11.69 dB (BAQ), 8.00 dB (FBAQ) and 11.94 dB (BAVQ) for E-SAR data and 8.77 dB (BAQ), 5.17 dB (FFT-BAQ) and 9.56 dB (BAVQ) for ERS1 data for a data resolution of 2 bits/sample were achieved with a reduction factor of about 3 for the E-SAR data and 2.5 for ERS1 data.