Evaluating the Effect of Stage Removal on the Performance of Phase Gradient Autofocus (PGA) Algorithm

Autofocus is a crucial step for Synthetic Aperture Radar (SAR) image formation. The choice of autofocus strategy is highly dependent on the compromise between SAR processing speediness and image quality demanded for trustworthy target analysis applications. The Phase Gradient Autofocus (PGA) algorithm for SAR phase error correction was released at the end of the 1980s. Since then, it has been evinced as a vigorous technique that can afford outstanding results over an assortment of SAR data not only on scene content but also for the hierarchy of the degrading phase error. Undeniably, PGA is adapted to restore defocused SAR images to their designed quality. The accomplishment of the algorithm is entrenched on four elemental signal processing steps, which comprise circular shifting, windowing, estimation of phase gradient, and lastly iterative correction. The question has been highlighted, which of these four elements is verily decisive to the efficacy of the algorithm. Breaking into the algorithm to dispense with one of these four steps reduces the computational cost and processing time while keeping algorithm performance and not affecting the image quality. The Breaking into the algorithm and the deep understanding of these four steps, the role of each part, and the importance of its implementation according to its arrangement within the algorithm is an attempt to reduce the computational complexity by reducing the steps of PGA which, may allow the creation of a customized algorithm to take the advantages of the outstanding PGA and overcome its slow implementation, This enables us to reduce the computational cost and processing time while keeping algorithm performance and not affecting the image quality using the algorithm in real time and without human intervention. Decomposition will be executed in three phases by masking one of the basic algorithm components in a cascaded manner to assess the effect on image quality, then elect which step may be eliminated while maintaining the image quality at an acknowledged level. The results showed the importance of implementing all four fundamental algorithm steps to achieve an acceptable level of image quality.