Font Size: a A A

Study On Three-dimensional Imaging Method Of Cylinder SAR For Security Inspection

Posted on:2021-07-31Degree:MasterType:Thesis
Country:ChinaCandidate:Y D CaoFull Text:PDF
GTID:2518306050955029Subject:Master of Engineering
Abstract/Summary:PDF Full Text Request
Synthetic Aperture Radar(SAR)is a kind of technology developed based on radar systems,which has been widely used in military and civilian fields.Since the beginning of the new century,the security of public is often threatened by terrorism.Therefore,security inspection equipment is required to ensure safety in areas with large numbers of people.The signal emitted by the millimeter-wave radar has a short wavelength and can penetrate human clothing without ionizing radiation.It is widely used in security inspection application.The three-dimensional imaging technology has become a key research content in the field of security inspection.It needs to have some features,such as obtaining comprehensive information about the target,high resolution,accurate compensation,and fast imaging.The three-dimensional imaging system needs to perform two-dimensional scanning for the scene,which increases the number of transmitting and receiving antennas,making the cost of hardware higher.It is necessary to transmit a wide-band signal for obtaining a high range resolution,which makes it challenging to develop an imaging system.In height and azimuth direction,a large synthetic aperture needs to form for achieving high resolution,so that the scanning method with narrow beam is no longer applicable.For practical applications,traditional frequency-domain algorithms such as RD and CS algorithms are no longer applicable for security inspection imaging,So the method of no approximate signal processing should be used to for imaging.Also,data processing has a specific challenge because of the real-time characteristic of security imaging.With SAR imaging as a technical background,this thesis uses a two-dimensional synthetic aperture scanning mode in cylinder SAR.Three-dimensional image reconstruction algorithms for the detection of scenes is proposed for cylinder SAR,and a high-speed data processing scheme based on CUDA platform is designed.Firstly,to obtain the information of the all-round scene and achieve high resolution of the three-dimensional image for security inspection,a SAR imaging system for cylinder SAR is studied in this thesis.In this system,the scanning mode of the wide beam stripe mode is adopted in height direction,and the beam scanning mode of the large-angle spotlight scanning is used in azimuth direction.Over the height-range synthetic aperture,the cylinder scanning is achieved,obtaining the high-resolution for imaging.A single-cylinder scan can obtain all-around information of the inspected person.A method of single transmission and double reception for signal is used,which saves the number of system antennas and reduces the cost of developing system development.Secondly,a three-dimensional imaging algorithm that combines the range migration algorithm(RMA)and back projection(BP)is studied in this thesis,solving the problem of inaccurate compensation of three-dimensional data.The proposed algorithm is used to process the sampled data from the scene in the cylinder,and it can realize the range migration correction without approximation.Then the accurate compensation and focus for image are completed.Besides,the algorithm is also suitable for non-ideal cylinder SAR imaging.Thirdly,to solve the problem of detecting only one person at a time in traditional security inspection,an imaging mechanism for the scene outside the cylinder is studied in this thesis.And an improved cylinder three-dimensional RMA imaging algorithm is proposed.Scanning a larger scene outside the cylinder can detect multiple people at the same time.A improved three-dimensional cylinder RMA imaging algorithm achieves image focus over two two-dimensional decoupling processes.It reduces the amount of calculation,having more robust engineering applicability.Finally,a parallel processing scheme is studied in this thesis,over combining CPU and GPU with CUDA platform.The computing efficiency is improved using asynchronous operations,stream processing,and multi-threaded parallel processing.The scheme meets the high speed processing of complex calculations,realizing the real-time processing for security inspections.
Keywords/Search Tags:Cylinder SAR, 3D Imaging, RMA, BP, Parallel Processing
PDF Full Text Request
Related items