Leier, Stefan (2014)
Signal Processing Techniques for Seafloor Ground-Range Imaging Using Synthetic Aperture Sonar Systems.
Technische Universität Darmstadt
Ph.D. Thesis, Primary publication
|
Text
Doktorarbeit_Leier_pub_v1.pdf Copyright Information: CC BY-NC-ND 2.5 Generic - Creative Commons, Attribution, NonCommercial, NoDerivs . Download (12MB) | Preview |
Item Type: | Ph.D. Thesis | ||||
---|---|---|---|---|---|
Type of entry: | Primary publication | ||||
Title: | Signal Processing Techniques for Seafloor Ground-Range Imaging Using Synthetic Aperture Sonar Systems | ||||
Language: | English | ||||
Referees: | Zoubir, Prof. Abdelhak M. ; Pesavento, Prof. Marius ; Andy, Prof. Schürr ; Franko, Prof. Küppers | ||||
Date: | 1 September 2014 | ||||
Place of Publication: | Darmstadt | ||||
Date of oral examination: | 18 July 2014 | ||||
Abstract: | In this Ph.D. thesis advanced signal processing techniques are addressed in order to reconstruct high-resolution seafloor imagery using synthetic aperture sonar ground- range imaging. This enables applications such as object detection, hydrography, and pipeline inspection, among others. In particular, the problems of echo-data-driven motion estimation known as micronavigation, and compensation of phase errors are considered. Based on the developed processing chain, a sensitivity study is conducted that points out the impact of distorted seafloor imagery on an automatic detection and classification system for target recognition. Furthermore, the framework of compressive sensing is introduced for synthetic aperture imaging to attain higher coverage rates using along-track undersampling. Synthetic aperture techniques are advantageous over conventional real aperture imaging techniques as they achieve a range-independent resolution that enables the reconstruction of high-quality sonar imagery. However, this requires an accurate knowledge of the positions of the transmitter and the receiving elements for multiple consecutive transmission and reception times. As inertial navigation systems are imprecise in tracking translational platform motion, echo-data-driven motion estimation is additionally employed to estimate the platform trajectory. For this purpose, a motion estimation processing chain for real sonar measurements is designed in this thesis, which considers height information about the seafloor. To this end, a topography estimation technique, which takes into account a continuous roll movement of the imaging platform, is developed. In order to avoid image defocus in environments with a strongly varying topography in case of nonlinear trajectories, the obtained height estimates are used during image reconstruction. This leads to significant image quality improvements, which is demonstrated on synthetic data. A comparison of real synthetic aperture sonar images then highlights the quality enhancement using the developed processing chain. Furthermore, practical methods for an unbiased estimation of platform motion are proposed. These involve a compensation technique for correcting the occurrence of biased time delay estimates due to a high ratio of carrier frequency to bandwidth as well as a broadside beamforming method that equalizes the varying time delays due to near-field scenarios and widebeam systems. In order to avoid image defocusing in the presence of residual phase errors, modifications of an existing autofocus technique are proposed to better cope with spatially varying point spread functions in stripmap synthetic aperture imaging. Additionally, a data-driven calibration method is developed so as to correctly estimate an optimal average sound speed yielding the best focused imagery in situations of a spatially varying sound-speed profile. Although data-driven motion estimation and phase error compensation techniques aim at achieving high-quality imagery of the seafloor, investigating the influence of distorted sonar imagery on automatic target recognition systems is of utmost importance for future autonomous sonar systems in order to judge their reliability. A sensitivity study is conducted that demonstrates significant performance loss in image segmentation, feature quality as well as in classification performance of a specific automatic detection and classification system. Further, an empirical relation between image degradation and performance loss of the individual stages of the automatic detection and classification system is highlighted. In order to guarantee reliable inputs for automatic target recognition, a strategy is proposed to sequentially assess the image quality of synthetic aperture sonar imagery during reconstruction, which is based on the instantaneous cross-range resolution. Increasing the coverage rate of a synthetic aperture system in the case of conventional imaging techniques is only feasible by increasing the physical array size. Alternatively, a compressive sensing framework is applied to perform aperture undersampling and, thus, offer a higher platform speed while still maintaining imaging performance. A stripmap imaging technique is developed to avoid the occurrence of azimuth image ambiguities. Synthetic data simulations then demonstrate the huge potential in data reduction, and laboratory experiments using compressive sensing further show an increase in platform speed by a factor of two. This possibly reduces the overall mission time of future synthetic aperture systems in real-life scenarios. All developed methods are applied to synthetic data as well as to real sonar measurements. |
||||
Alternative Abstract: |
|
||||
URN: | urn:nbn:de:tuda-tuprints-40868 | ||||
Classification DDC: | 600 Technology, medicine, applied sciences > 620 Engineering and machine engineering | ||||
Divisions: | 18 Department of Electrical Engineering and Information Technology > Institute for Telecommunications > Signal Processing | ||||
Date Deposited: | 01 Sep 2014 07:26 | ||||
Last Modified: | 25 Jan 2024 10:11 | ||||
URI: | https://tuprints.ulb.tu-darmstadt.de/id/eprint/4086 | ||||
PPN: | 386756570 | ||||
Export: |
View Item |