Microwave Imaging Algorithms Based on Genetic Algorithm

In this thesis, the GA-based reconstruction algorithms were optimized by various techniques, the choice of the best setting, acceleration by custom hardware, and by taking advantages of the problem nature (i.e., if a high contrast exists). Those techniques are applied for 1D and 2D only. We showed that the choice of the appropriate settings for the GA parameters for 1D inverse scattering problem is of a critical impact on performance measures. Also, with the communication overheads are avoided – by implementing GA algorithm itself in the FPGA – the overall performance of our hardware accelerator for 1D problem can reach 240x speeding up when compared to PC(S/W only) version. Finally, our novel intelligent-decision GA-based technique reduces the time of computation by a factor of 50 while achieving a more accurate reconstructed image.