Synthetic Aperture Radar (SAR) is one of the most common imaging methods that can provide good quality images in any weather conditions where even optical systems are unable to function properly. This radar, due to its installation on a moving platform, creates a so-called large artificial aperture with its movement an‎d forms a high-resolution an‎d high-quality image by calculating the radar cross-section of the targets. Despite the advantages of this radar, very large volume of raw data has always been a major challenge in SAR processing, an‎d any effort an‎d innovation to reduce the volume of this data an‎d facilitate its processing will be valuable. One of the main ideas to achieve this goal is the use of compressed sensing (CS) theory, which, after its introduction, has also received much attention for use in SAR imaging applications. Using this technique, it is possible to sample the input signal at a rate much lower than the Nyquist rate in some applications such as SAR that show sparsity property, which caused a major revolution in many areas of signal processing. In imaging applications, fast methods such as OMP an‎d IHT were initially used, which in addition to drastically reducing the volume of raw data, also had high execution speed. But as the volume of data increased, researchers continued to look for ways to further reduce the volume of this data. One of the ideas proposed was to reduce the bit width of the input samples, which ultimately led to extensive research in the field of compressed sensing with one-bit samples. In addition to reducing the volume of data, compressed sensing with one-bit samples greatly simplifies the structure of the radar sampler hardware an‎d greatly reduces its cost. One of the most famous of these methods is called BIHT, which has also been used in imaging applications. In these methods, the amplitude information is completely lost an‎d leads to a decrease in the quality of the final image, but in radars such as SAR that have a sparse block property, block wise one-bit compressed sensing methods such as PBIBHT perform better in signal reconstruction compared to simple methods such as BIHT. In this thesis, while examining various methods of one-bit compressed sensing, four new ideas for block-based one-bit compressed sensing are suggested. The first algorithm, called Entropy based Binary Iterative Block Hard Thresholding (EBIBHT), has a cost function which is different from other methods an‎d it shows better performance in different scenarios compared to its competitor, PBIBHT. The second one is the block version of FISTA algorithm which is called Block-FISTA. The third one is the block version of RBIHT algorithm which is called Block-RBIHT, an‎d the last one is the block version of firm thresholding which can be applied to all of existing block-wise one-bit recovery algorithms. Finally, we will compare the performance of these methods on synthetic an‎d also real SAR data.

محمود مدرس هاشمي

محمدرضا تابان , محمدمهدي نقش