A Study On Recursive Backprojection Imaging Algorithm For Video SAR And Its DSP Implementation

Video Synthetic Aperture Radar(Video SAR or ViSAR,for short)can provide real-time dynamic monitoring in all weather at a high frame rate and has widely been used in the military and civilian fields.Typical imaging processing algorithms can be grouped into two categories: Polar Format Algorithm(PFA,for short)and Backprojection(BP,for short)algorithm.Among them,the BP algorithm can perform sequential processing,which is very suitable to process the Video SAR data.However,it has very high requirements on the computation capability and storage space.Therefore,a recursive BP algorithm has been developed,which can greatly save the storage space for overlapping aperture processing,and greatly improve the computing efficiency at a higher aperture overlap ratio.However,the recursive BP algorithm has critical requirements on the floating-point computation capability and storage capability of the hardware when the Video SAR works in large scenes.In order to meet the requirements on the real-time,low-power,low-cost,and lightweight for Video SAR systems,the design of the embedded imaging processing module of the system is particularly important.Compared with other microprocessors,dedicated Digital Signal Processors(DSP,for short)have the capability of tens billions of floating-point operations per second,large-capacity data access and abundant external interface,which is especially suitable to an embedded hardware platform for the Video SAR imaging processing module.In the real-world applications,it is an urgent task to combine recursive BP algorithm with the DSP hardware to produce high-resolution real-time dynamic frame images.This thesis mainly studies the Video SAR imaging processing algorithm and its implementation technique on the DSP hardware platform.The main accomplishments are summarized as follows:1.Investigated and compared several representative algorithms developed for the Video SAR imaging processing in detail.Specifically,we have analyzed the principles of the PFA,BP and recursive BP algorithms and compared them through computer simulations in different scenarios.The pros and cons of these algorithms are revealed based on the simulation results.In the final,the recursive BP algorithm is selected for the further technical verification and analysis on the DSP hardware platform.2.Thoroughly studied the DSP chip TMS320C6678 and its hardware platform.This thesis explores the applicability and efficiency of the DSP chip in the field of Video SAR imaging from the perspectives of chip selection,architecture,SYS/BIOS programming method,and multi-core parallel mode.3.Designed the parallel architecture for the recursive BP algorithm based on the DSP chip TMS320C6678.This thesis explores the implementation and optimization procedures of the recursive BP algorithm with the DSP chip in detail,and expounds the important roles of SYS/BIOS architecture configuration,architecture design and Cache consistency in the implementation procedure of the algorithm.The effects of the “for loop”,“pragmas” and keywords,built-in functions,and optimization level on the performance of the algorithm optimization are discussed.4.Verified the recursive BP algorithm on the DSP platform by using the Gotcha data.By analyzing the characteristics of the corner reflectors and various targets in the Gotcha imaging results,this thesis equivalently verified the feasibility of the recursive BP algorithm on the DSP platform.In this thesis,we have successfully realized the recursive BP algorithm on the DSP hardware platform.On a single-chip 8-core DSP platform,a frame of image with a size of 512×512pixels and imaging resolution of 0.3 m is obtained under the third-order iteration.It provides the theoretical basis and a realization idea for the real-time imaging processing in Video SAR.