摘要

Fast factorized back-projection (FFBP) is an effective algorithm constructed based on aperture factorized and image recursive fusion. It is shown that the number of operations can be made comparable to frequency domain algorithms and the image quality is barely sacrificed compared with back-projection (BP). Different from frequency domain algorithms, rectangular coordinate system images and polar coordinate system images reconstructed by FFBP cannot satisfy the requirement of traditional autofocus. To solve this problem, we establish a pseudo-polar coordinate system for reconstruction in FFBP firstly, which paves a way for autofocusing method. Secondly, multi-aperture operations are not only performed in FFBP but also are used in multiple aperture mapdrift (MAM), This characteristic becomes a great advantage to incorporate MAM into FFBP iterations seemly and achieve the success of data based motion compensation. Finally, the results of collected data sets validate the feasibility and effectiveness of the proposed approach.

全文