WO2018064917A1 - 基于gpu的人体微波回波模拟方法及*** - Google Patents

基于gpu的人体微波回波模拟方法及*** Download PDF

Info

Publication number
WO2018064917A1
WO2018064917A1 PCT/CN2017/096100 CN2017096100W WO2018064917A1 WO 2018064917 A1 WO2018064917 A1 WO 2018064917A1 CN 2017096100 W CN2017096100 W CN 2017096100W WO 2018064917 A1 WO2018064917 A1 WO 2018064917A1
Authority
WO
WIPO (PCT)
Prior art keywords
gpu
human body
echo
thread
phase center
Prior art date
Application number
PCT/CN2017/096100
Other languages
English (en)
French (fr)
Inventor
祁春超
王爱先
陈寒江
黄雄伟
刘艳丽
贾成艳
赵术开
Original Assignee
华讯方舟科技有限公司
深圳市太赫兹科技创新研究院
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华讯方舟科技有限公司, 深圳市太赫兹科技创新研究院 filed Critical 华讯方舟科技有限公司
Priority to US16/340,332 priority Critical patent/US11550078B2/en
Publication of WO2018064917A1 publication Critical patent/WO2018064917A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01VGEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V13/00Manufacturing, calibrating, cleaning, or repairing instruments or devices covered by groups G01V1/00 – G01V11/00
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01VGEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V3/00Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
    • G01V3/12Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation operating with electromagnetic waves
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01VGEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V8/00Prospecting or detecting by optical means
    • G01V8/005Prospecting or detecting by optical means operating with millimetre waves, e.g. measuring the black losey radiation
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01VGEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V8/00Prospecting or detecting by optical means
    • G01V8/10Detecting, e.g. by using light barriers
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/004Artificial life, i.e. computing arrangements simulating life

Definitions

  • the invention belongs to the field of echo simulation technology, and in particular relates to a human body microwave echo simulation method and system based on a GPU (Graphics Processing Unit).
  • GPU Graphics Processing Unit
  • the object of the embodiments of the present invention is to provide a GPU-based human body microwave echo simulation method and system, aiming at solving the above-mentioned echo dimension of the human body three-dimensional scanning imaging system.
  • the embodiment of the present invention is implemented by using a GPU-based human body microwave echo simulation method, and the method includes:
  • the kernel function calculates the simulated echo of the human body microwave in parallel by running in parallel in the multi-threaded manner in the plurality of processing cores of the GPU device according to the parallel computing network parameter;
  • the GPU end transmits the calculated analog echo of the human body microwave to the GPU host.
  • Another object of the embodiments of the present invention is to provide a GPU-based human body microwave echo simulation system, including a CPU host end and a GPU device end, where the CPU host end includes a simulation parameter sending unit, a parallel parameter setting unit, and a kernel function start. Unit, the GPU device side includes a parallel computing unit and an echo data return unit, wherein:
  • the simulation parameter sending unit is configured to transfer the simulation input parameter from the memory of the CPU host end to the video memory of the GPU device end;
  • the parallel parameter setting unit is configured to set a parallel computing network parameter running on the GPU device end on the CPU host side;
  • a kernel function starting unit configured to start a kernel function preset for the microwave echo simulation of the human body in the CPU host end
  • the parallel computing unit is configured to calculate, in the multi-threaded manner, the analog echo of the human body microwave in a plurality of processing cores of the GPU device according to the parallel computing network parameter under the control of the kernel function;
  • the echo data returning unit is configured to send back an analog echo of the human body microwave calculated by the GPU device to the GPU host.
  • the simulation input parameter is transmitted from the memory of the CPU host to the memory of the GPU device; the parallel computing network parameter running on the GPU device is set on the CPU host; and the CPU host is preset.
  • a kernel function for human body microwave echo simulation is operated in parallel in a plurality of processing cores of the GPU device according to the parallel computing network parameter to calculate an analog echo of the human body microwave
  • the GPU end transmits back the simulated echo of the human body microwave to the GPU host end, thereby fully utilizing the characteristics that the GPU can perform parallel computing, accelerating the echo simulation process, and greatly improving the human body microwave scanning imaging.
  • the real-time nature of the system echo simulation is transmitted from the memory of the CPU host to the memory of the GPU device; the parallel computing network parameter running on the GPU device is set on the CPU host; and the CPU host is preset.
  • a kernel function for human body microwave echo simulation is operated in parallel in a plurality of processing cores of the GPU device according to the parallel computing network parameter to calculate an analog echo of the human body microwave
  • the GPU end transmits
  • FIG. 2 is a schematic structural diagram of a GPU-based parallel computing using a GPU in a GPU-based human body microwave echo simulation method according to an embodiment of the present invention
  • FIG. 3 is a schematic flowchart of a GPU-based human body microwave echo simulation method according to an embodiment of the present invention
  • step S304 is a schematic flowchart of step S304 in a GPU-based human body microwave echo simulation method according to an embodiment of the present invention
  • FIG. 5 is a schematic block diagram of a GPU-based human body microwave echo simulation system according to an embodiment of the present invention.
  • the embodiment of the invention is for realizing the three-dimensional echo simulation of the human body in the three-dimensional scanning imaging security system of the short-distance human body, and the three-dimensional scanning imaging security system of the short-distance human body can be divided into the plane scanning according to the scanning surface formed by the phase center of the transmitting and receiving antenna unit.
  • the surface system and the cylindrical scanning system are mainly for the case of the cylindrical scanning geometry in the embodiment of the present invention, but the echo simulation method provided by the present invention can also be applied to the planar scanning geometry.
  • FIG. 1 is a geometrical view showing a cylindrical scan of a human body three-dimensional scanning imaging security inspection system in an embodiment of the present invention.
  • the transmitting and receiving antenna units form a linear array in the vertical height Z direction, and the distance from the central axis of the cylindrical scanning surface assembly is ⁇ .
  • the linear array is also rotated and scanned, and finally formed on the cylinder surface.
  • the equivalent phase center of multiple antennas such as anti-counterfeiting interval and equal height coordinate.
  • Cylindrical coordinates Indicates the position of the phase center of an antenna, where The azimuth coordinate value of the sampling point in the mth azimuth direction under the azimuth direction equal interval sampling condition, the range of m is 0 ⁇ m ⁇ M-1, where M represents the total number of points of the antenna phase center scanning along the azimuth direction; z n represents the height coordinate value of the nth height to the sampling point in the height direction equal interval sampling condition, and the value range of n is 0 ⁇ n ⁇ N-1, where N represents the total number of points of the antenna phase center scanning in the height direction.
  • the human body microwave three-dimensional echo simulation is to calculate the phase center of all antennas by means of data simulation. Echo data at the place.
  • the echo simulation method provided by the embodiment of the present invention is based on a GPU. Therefore, before describing the details of the embodiments of the present invention in detail, some concepts and steps of parallel computing using the GPU are briefly described.
  • the GPU integrates a large number of computing units on a limited number of boards, and is mainly responsible for computing high-density computing tasks. Compared with CPUs in conventional computers or workstations, GPUs have stronger floating-point computing power and larger storage bandwidth.
  • the GPU is used as the host side, and the GPU is used as a device side for processing, so that the GPU can run some programs that can be highly threaded.
  • the highly threaded program constitutes a so-called kernel function that is executed in parallel on the GPU.
  • the kernel function runs in the operation mode of the CPU in the organization of the thread block. Up, and each thread block contains multiple threads executing in parallel.
  • parallel computing network parameters need to be set according to the needs of parallel optimization, including the dimension of the thread block for parallel computing on the GPU device side and the dimension of the thread within one thread block.
  • FIG. 3 is a GPU-based human body microwave echo simulation according to an embodiment of the present invention. Schematic flow chart of the method. Referring to FIG. 3, the method may include the following steps:
  • the simulation input parameters are transferred from the memory of the CPU host to the video memory of the GPU device.
  • the simulation input parameters include: scan parameters of the human body scanning imaging system, antenna phase center position parameters, and human body model distributed target parameters.
  • the scan parameters of the human body scanning imaging system include a transmit pulse signal center frequency f c , a pulse width T S , a bandwidth B, and an echo sampling frequency f s .
  • the antenna phase center position parameter is ⁇ represents the distance between the phase center of the antenna of the human body microwave security system and the central axis of the cylinder under the cylindrical scanning geometry;
  • z n represents the height coordinate value of the nth height to the sampling point in the height direction equal interval sampling condition, and the value range of n is 0 ⁇ n ⁇ N-1, where N represents the total number of points of the antenna phase center scanning in the height direction.
  • the human body model distributed target parameters Represents the distance, azimuth, and height coordinates of the kth scattering point on the human body model, Indicates the magnitude of the scattering coefficient of the kth scattering point on the human body model.
  • the value of k ranges from 0 ⁇ k ⁇ K-1, where K represents the total number of scattering points in the distributed target of the human body model.
  • the parallel computing network parameter includes a dimension of a thread block in which the GPU device side performs parallel computing and a dimension of a thread in each thread block.
  • the dimension of the thread block is M ⁇ N, where M represents the total number of points of the antenna phase center scanned in the azimuth direction, and N represents the total number of points of the antenna phase center scanned in the height direction, and the dimension in the one thread block It is K ⁇ 1, where K represents the total number of scattering points in the distributed target of the human body model.
  • the kernel function calculates the simulated echo of the human body microwave in parallel by running in parallel in the multi-threaded manner in the plurality of processing cores of the GPU device according to the parallel computing network parameter.
  • FIG. 4 shows a specific implementation process of step S304.
  • the specific implementation process of step S304 is as follows:
  • the antenna phase center position commonly used by all threads in the thread block is determined according to the number of the thread block in which the current GPU thread is located in all the thread blocks.
  • the number of the thread block in which the current GPU thread is located is (m', n'), and the antenna phase center position used by all the threads in the current thread block is determined to be
  • represents the distance from the center axis of the antenna of the human body microwave security system under the cylindrical scanning geometry to the central axis of the cylinder;
  • z n' represents the height coordinate value of the nthth height sampling point under the condition of equal interval sampling, and the range of n' is 0 ⁇ n' ⁇ N-1, where N represents the antenna phase center
  • N represents the antenna phase center
  • the position coordinates and the scattering coefficient size of the mannequin distributed target scattering point used in the current GPU thread calculation are determined according to the number of the current GPU thread in the thread block to which it belongs.
  • another current GPU is located in the thread block to which it belongs, and the position coordinate of the distributed target scattering point of the k'th body model used in the current thread calculation is determined to be
  • the scattering coefficient is Where k' is in the range 0 ⁇ k' ⁇ K-1, where K represents the total number of scatter points in the distributed target of the human body model.
  • a distance gate unit position in which the delay of the human distributed target scattering point to the antenna phase center position is located in the echo is determined.
  • the time delay unit ⁇ d of the k'th scattering point to the center of the antenna phase is in the unit position of the distance gate in the echo.
  • f s is the echo sampling frequency
  • the order is located
  • the echo data received at the antenna phase center is Then there are:
  • the echo signal data is convoluted with the transmitted signal pulse of the human body scanning imaging system to obtain finally outputted echo signal data using the transmitted signal pulse.
  • the emission signal pulse of the human body scanning imaging system is a chirp signal pulse, and the expression is:
  • t denotes the time coordinate
  • f c , T S and B denote the human body microwave of the simulation input, the center frequency, the pulse width and the bandwidth parameter of the three-dimensional imaging security system, respectively.
  • the other thread blocks of the GPU device end perform the above steps S401 to S407 in parallel to obtain the center position of all the antennas.
  • Echo signal data Where 0 ⁇ m ⁇ M-1; 0 ⁇ n ⁇ N-1.
  • the GPU end transmits back the simulated echo of the calculated human body microwave to the GPU host.
  • the CPU host sends the analog echo to the subsequent imaging processing flow after receiving the simulated echo of the human body microwave, or saves the simulated echo data to the hard disk according to the storage format and the path requirement. .
  • the GPU-based human body microwave echo simulation method is provided by transferring the simulation input parameters from the memory of the CPU host to the memory of the GPU device; and setting the operation at the CPU host.
  • the analog echo of the human body microwave is calculated by running in parallel in the processing cores of the device; the GPU end transmits the calculated simulated echo of the human body microwave to the GPU host, so that the GPU can be fully utilized.
  • the characteristics of parallel computing accelerates the echo simulation process, greatly improving the real-time performance of the echo simulation of the human microwave scanning imaging system.
  • the GPU is used instead of the CPU for echo simulation, CPU resources can be saved for system control and resource scheduling.
  • logic analysis and other functions in addition, the echo simulation can change the simulation parameters as needed, so that the entire echo simulation Easy-to-use help system development and debugging a high efficiency, and reduce costs.
  • FIG. 5 is a schematic block diagram of a GPU-based human body microwave echo simulation system according to an embodiment of the present invention, which is used to operate the method provided by the embodiment shown in FIG. 3 to FIG. 4 of the present invention. Only the parts related to the present embodiment are shown for convenience of explanation.
  • a GPU-based human body microwave echo simulation system includes a CPU host end 1 and a GPU device end 2, and the CPU host end 1 includes a simulation parameter sending unit 12 and parallel parameter settings.
  • the unit 11 and the kernel function starting unit 13 the GPU device terminal 2 includes a parallel computing unit 21 and an echo data return unit 22, wherein:
  • the simulation parameter sending unit 12 is configured to transfer the simulation input parameter from the memory of the CPU host 1 to the video memory of the GPU device 2;
  • the parallel parameter setting unit 11 is configured to set a parallel computing network parameter running on the GPU device end 2 at the CPU host end 1;
  • the kernel function starting unit 13 is configured to start a kernel function preset for the microwave echo simulation of the human body in the CPU host end 1;
  • the parallel computing unit 21 is configured to perform parallel simulation calculation of the human body microwave in a plurality of processing cores of the GPU device terminal 2 in a multi-threaded manner according to the parallel computing network parameter under the control of the kernel function. Echo
  • the echo data returning unit 22 is configured to send back the simulated echo of the human body microwave calculated by the GPU device terminal 2 to the GPU host.
  • the simulation input parameters include: scan parameters of the human body scanning imaging system, antenna phase center position parameters, and human body model distributed target parameters.
  • the parallel computing network parameter includes a dimension of a thread block in which the GPU device end 2 performs parallel computing and a dimension of a thread in each thread block.
  • the parallel computing unit 21 is specifically configured to:
  • the transmit signal pulse of the human body scanning imaging system is a chirp signal pulse.
  • the GPU-based human body microwave echo simulation system provided by the embodiment of the present invention can also fully utilize the feature that the GPU can perform parallel computing, accelerate the echo simulation process, and greatly improve the echo of the human body microwave scanning imaging system.
  • Real-time simulation in addition, because the GPU is used instead of the CPU for echo simulation, CPU resources can be saved for system control, resource scheduling, logic analysis, etc.
  • the simulation parameters can be changed as needed during echo simulation.
  • the echo simulation process is easy to use and helps to improve the efficiency and cost of a high system.

Landscapes

  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • Geophysics (AREA)
  • Manufacturing & Machinery (AREA)
  • Theoretical Computer Science (AREA)
  • Remote Sensing (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Geology (AREA)
  • Environmental & Geological Engineering (AREA)
  • Electromagnetism (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

一种基于GPU的人体微波回波模拟方法及***,其中方法包括:将仿真输入参数由中央处理单元CPU主机端(1)的内存传送至图像处理单元GPU设备端(2)的显存中(S301);在CPU主机端(1)设置运行在GPU设备端(2)的并行计算网络参数(S302);启动CPU主机端(1)内预置的用于人体微波回波模拟的内核函数(S303);内核函数根据并行计算网络参数以多线程的方式在GPU设备端(2)的多个处理核中并行运行计算得出人体微波的模拟回波(S304);GPU设备端将(2)计算得出的人体微波的模拟回波回传至CPU主机端(1)(S305)。本方案可以充分利用GPU可进行并行计算的特点,加速回波模拟过程,大大提高了人体微波扫描成像***回波模拟的实时性。

Description

基于GPU的人体微波回波模拟方法及*** 技术领域
本发明属于回波模拟技术领域,尤其涉及一种基于GPU(Graphics Processing Unit,图形处理单元)的人体微波回波模拟方法及***。
背景技术
基于微波、毫米波或者太赫兹波的近距离人体三维扫描成像安检***因其能够穿透人体衣物,对人体辐射剂量小,能够识别藏匿于人体周身的金属或非金属违禁品等优点,在近些年来受民众关注,并且受到部分机场、法院、监狱等需要安检的组织机构的推崇。在人体三维扫描成像安检***的研制和测试阶段,通常需要进行回波模拟仿真生成,以便支撑***参数验证分析、***成像机理研究、成像算法设计与验证等环节。而由于人体三维扫描成像***的回波维度较高,***中通道较多,导致回波模拟仿真时计算量大、复杂度高,因此通常无法使用当前常规计算机或者工作站中的CPU(Central Processing Unit,中央处理单元)来实时计算生成仿真回波。
发明内容
本发明实施例的目的在于提供一种基于GPU的人体微波回波模拟方法及***,旨在解决上述由于人体三维扫描成像***的回波维度 较高,***中通道较多,导致回波模拟仿真时计算量大、复杂度高,因此通常无法使用当前常规计算机或者工作站中的CPU来实时计算生成仿真回波的问题。
本发明实施例是这样实现的,一种基于GPU的人体微波回波模拟方法,该方法包括:
将仿真输入参数由CPU主机端的内存传送至GPU设备端的显存中;
在所述CPU主机端设置运行在所述GPU设备端的并行计算网络参数;
启动所述CPU主机端内预置的用于人体微波回波模拟的内核函数;
所述内核函数根据所述并行计算网络参数以多线程的方式在所述GPU设备端的多个处理核中并行运行计算得出人体微波的模拟回波;
所述GPU端将计算得出的人体微波的模拟回波回传至所述GPU主机端。
本发明实施例的另一目的在于提供一种基于GPU的人体微波回波模拟***,包括CPU主机端和GPU设备端,所述CPU主机端包括仿真参数发送单元、并行参数设置单元以及内核函数启动单元,所述GPU设备端包括并行计算单元和回波数据返回单元,其中:
所述仿真参数发送单元,用于将仿真输入参数由CPU主机端的内存传送至GPU设备端的显存中;
所述并行参数设置单元,用于在所述CPU主机端设置运行在所述GPU设备端的并行计算网络参数;
内核函数启动单元,用于启动所述CPU主机端内预置的用于人体微波回波模拟的内核函数;
所述并行计算单元,用于在所述内核函数控制下根据所述并行计算网络参数以多线程的方式在所述GPU设备端的多个处理核中并行运行计算得出人体微波的模拟回波;
所述回波数据返回单元,用于将所述GPU设备端计算得出的人体微波的模拟回波回传至所述GPU主机端。
实施本发明实施例提供的一种基于GPU的人体微波回波模拟方法及***具有以下有益效果:
本发明实施例通过将仿真输入参数由CPU主机端的内存传送至GPU设备端的显存中;在所述CPU主机端设置运行在所述GPU设备端的并行计算网络参数;启动所述CPU主机端内预置的用于人体微波回波模拟的内核函数;所述内核函数根据所述并行计算网络参数以多线程的方式在所述GPU设备端的多个处理核中并行运行计算得出人体微波的模拟回波;所述GPU端将计算得出的人体微波的模拟回波回传至所述GPU主机端,从而可以充分利用GPU可进行并行计算的特点,加速回波模拟过程,大大提高了人体微波扫描成像***回波模拟的实时性。
附图说明
图1是本发明实施例中人体三维扫描成像安检***的柱面扫描 几何示意图;
图2是本发明实施例提供的一种基于GPU的人体微波回波模拟方法中利用GPU进行并行计算的架构示意图;
图3是本发明实施例提供的一种基于GPU的人体微波回波模拟方法的示意流程图;
图4是本发明实施例提供的一种基于GPU的人体微波回波模拟方法中步骤S304的示意流程图;
图5是本发明实施例提供的一种基于GPU的人体微波回波模拟***的示意性框图。
具体实施方式
为了使本发明的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本发明进行进一步详细说明。应当理解,此处所描述的具体实施例仅仅用以解释本发明,并不用于限定本发明。
本发明实施例是针对近距离人体三维扫描成像安检***中实现人体微波三维回波模拟的,近距离人体三维扫描成像安检***根据发射接收天线单元的相位中心所形成的扫描面可分为平面扫面***和柱面扫描***,本发明实施例中主要是针对柱面扫几何的情况,但本发明提供的回波模拟方法同样可用于平面扫描几何情况。
图1示出了本发明实施例中人体三维扫描成像安检***的柱面扫描的几何示意图。参见图1所示,收发天线单元在垂直高度Z方向组成线性阵列,距柱面扫面集合中心轴线距离为ρ,同时,该线性阵列还经旋转扫描,最后等效形成在柱面上分布的等防伪间隔、等高度 向坐标的多个天线等效相位中心。以柱面坐标
Figure PCTCN2017096100-appb-000001
表示某个天线相位中心的位置,其中
Figure PCTCN2017096100-appb-000002
表示方位角方向等间隔采样条件下第m个方位角方向采样点的方位角坐标值,m的范围是0≤m≤M-1,其中M表示天线相位中心沿方位角方向扫描的总点数;zn表示高度方向等间隔采样条件下第n个高度向采样点的高度坐标值,n的取值范围是0≤n≤N-1,其中N表示天线相位中心沿高度方向扫描的总点数。人体微波三维回波模拟即是通过数据仿真的手段计算所有天线相位中心
Figure PCTCN2017096100-appb-000003
处的回波数据。
本发明实施例提供的回波模拟方法是基于GPU进行的,因此在详细介绍本发明的实施方式的细节之前,先简单描述利用GPU进行并行计算的一些概念和步骤。
GPU是在有限的板卡上集成大量的计算单元,主要负责计算密度高的运算任务,相比于常规计算机或工作站中的CPU,GPU具有更强的浮点数运算能力和更大的存储宽带。通常情况下,将GPU作为主机端,将GPU作为协助处理的设备端,从而让GPU来运行一些能够被高度线程化的程序。如图2所示,为利用GPU进行并行加速计算的架构示意图,高度线程化的程序组成一个所谓的内核函数被并行执行在GPU上,内核函数以线程块的组织方式运行在CPU的一个运算核心上,而每个线程块又包含多个并行执行的线程。在实际运行中,需根据并行优化的需要设置并行计算网络参数,包括GPU设备端进行并行计算的线程块的维度以及一个线程块内线程的维度。
图3是本发明实施例提供的一种基于GPU的人体微波回波模拟 方法的示意流程图。参见图3所示,该方法可以包括以下步骤:
在S301中,将仿真输入参数由CPU主机端的内存传送至GPU设备端的显存中。
在本实施例中,所述仿真输入参数包括:人体扫描成像***的扫描参数、天线相位中心位置参数以及人体模型分布式目标参数。
其中,所述人体扫描成像***的扫描参数包括发射脉冲信号中心频率fc,脉冲宽度TS、带宽B以及回波采样频率fs
其中,天线相位中心位置参数为
Figure PCTCN2017096100-appb-000004
ρ表示在柱面扫描几何下的人体微波安检***天线相位中心距柱面中轴线的距离大小;
Figure PCTCN2017096100-appb-000005
表示方位角方向等间隔采样条件下第m个方位角方向采样点的方位角坐标值,m的范围是0≤m≤M-1,其中M表示天线相位中心沿方位角方向扫描的总点数;zn表示高度方向等间隔采样条件下第n个高度向采样点的高度坐标值,n的取值范围是0≤n≤N-1,其中N表示天线相位中心沿高度方向扫描的总点数。
其中,人体模型分布式目标参数
Figure PCTCN2017096100-appb-000006
表示人体模型上第k个散射点的距离、方位角和高度坐标,
Figure PCTCN2017096100-appb-000007
表示人体模型上第k个散射点的散射系数大小,k的取值范围为0≤k≤K-1,其中K表示人体模型分布式目标中散射点的总个数。
在S302中,在所述CPU主机端设置运行在所述GPU设备端的并行计算网络参数。
在本实施例中,所述并行计算网络参数包括所述GPU设备端进行并行计算的线程块的维度以及每个线程块内线程的维度。
优选的,所述线程块的维度为M×N,其中M表示天线相位中心沿方位角方向扫描的总点数,N表示天线相位中心沿高度方向扫描的总点数,所述一个线程块内的维度是K×1,其中K表示人体模型分布式目标中散射点的总个数。
在S303中,启动所述CPU主机端内预置的用于人体微波回波模拟的内核函数。
在S304中,所述内核函数根据所述并行计算网络参数以多线程的方式在所述GPU设备端的多个处理核中并行运行计算得出人体微波的模拟回波。
图4示出了步骤S304的具体实现流程,参见图4所示,步骤S304的具体实现流程如下:
在S401中,根据当前GPU线程所在的线程块在所有线程块中所处的编号确定该线程块内所有线程共同使用的天线相位中心位置。
其中,令当前GPU线程所在的线程块在所述线程块中所处的编号为(m',n'),则确定当前线程块内所有线程共同使用的天线相位中心位置为
Figure PCTCN2017096100-appb-000008
其中ρ表示在柱面扫描几何下的人体微波安检***天线相位中心距柱面中轴线的距离大小;
Figure PCTCN2017096100-appb-000009
表示方位角方向等间隔采样条件下第m'个方位角方向采样点的方位角坐标值,m'的取值范围是0≤m'≤M-1,其中M表示天线相位中心沿方位角方向扫描的总点数;zn'表示高度向等间隔采样条件下第n'个高度采样点的高度坐标值,n'的取值范围是0≤n'≤N-1,其中N表示天线相位中心沿高度方向扫描的总点数。
在S402中,根据当前GPU线程在其所属线程块中所处的编号确定当前GPU线程计算所用到的人体模型分布式目标散射点的位置坐标和散射系数大小。
在本实施例中,另当前GPU在其所属线程块中所处的编号为k',那么确定当前线程计算所用到的第k'个人体模型分布式目标散射点的位置坐标为
Figure PCTCN2017096100-appb-000010
散射系数大小为
Figure PCTCN2017096100-appb-000011
其中k'的范围是0≤k'≤K-1,其中K表示人体模型分布式目标中散射点的总个数。
在S403中,计算所述天线相位中心位置和所述人体模型分布式目标散射点之间的距离,并根据所述距离计算出从所述人体模型分布式目标散射点到所述天线相位中心位置的时延。
在本实施例中,所述天线相位中心位置
Figure PCTCN2017096100-appb-000012
和第k'个散射点
Figure PCTCN2017096100-appb-000013
之间的距离大小为R,所述目标散射点所所述天线相位中心位置的时延为τd,那么有:τd=2R/c,其中c为电磁波在真空中的传播速度。
在S404中,确定所述人体分布式目标散射点到所述天线相位中心位置的时延在回波中所处的距离门单元位置。
在本实施例中,所述第k'个散射点到天线相位中心的时延τd在回波中所处的距离门单位位置即为
Figure PCTCN2017096100-appb-000014
其中fs为回波采样频率,
Figure PCTCN2017096100-appb-000015
为向下取整算子。
在S405中,将所述距离门单元内的回波信号加上来自所述人体分布式目标散射点的多普勒相位得到所述天线相位中心收到的所述距离门单元的回波信号数据。
在本实施例中,令位于
Figure PCTCN2017096100-appb-000016
处的天线相位中心收到的回波数据为
Figure PCTCN2017096100-appb-000017
那么有:
Figure PCTCN2017096100-appb-000018
在S406中,重复执行上述步骤S402~S405,直到当前线程块内所有的线程均执行完毕,得到所述天线相位中心位置收到的所有距离门单元的回波信号数据。
在本实施例中,重复执行上述步骤S402~S405,知道当前线程块内所有的线程均执行完毕,得到位于
Figure PCTCN2017096100-appb-000019
处的天线相位中心收到的所有距离门单元的回波数据
Figure PCTCN2017096100-appb-000020
在S407中,将所述回波信号数据与所述人体扫描成像***的发射信号脉冲进行卷积计算,得到最终输出的采用该发射信号脉冲的回波信号数据。
在本实施例中,所述人体扫描成像***的发射信号脉冲为线性调频信号脉冲,其表达式为:
Figure PCTCN2017096100-appb-000021
其中,t表示时间坐标,fc、TS和B分别表示仿真输入的人体微波、三维成像安检***的中心频率、脉冲宽度和带宽参数。
在S408中,在所述GPU设备端执行的其它线程块内并行执行上述步骤S401~步骤S407,最终得到对应所有天线相位中心位置的回波信号数据。
在本实施例中,在所述GPU设备端的其他线程块并行的执行上述步骤S401~步骤S407即可得到对应所有天线相位中心位置
Figure PCTCN2017096100-appb-000022
的回波信号数据
Figure PCTCN2017096100-appb-000023
其中,0≤m≤M-1;0≤n≤N-1。
在S305中,所述GPU端将计算得出的人体微波的模拟回波回传至所述GPU主机端。
在本实施例中,所述CPU主机端接收到所述人体微波的模拟回波后将所述模拟回波送往后续成像处理流程,或者根据存储格式和路径要求将模拟回波数据保存至硬盘。
以上可以看出,本实施例提供的一种基于GPU的人体微波回波模拟方法由于通过将仿真输入参数由CPU主机端的内存传送至GPU设备端的显存中;在所述CPU主机端设置运行在所述GPU设备端的并行计算网络参数;启动所述CPU主机端内预置的用于人体微波回波模拟的内核函数;所述内核函数根据所述并行计算网络参数以多线程的方式在所述GPU设备端的多个处理核中并行运行计算得出人体微波的模拟回波;所述GPU端将计算得出的人体微波的模拟回波回传至所述GPU主机端,从而可以充分利用GPU可进行并行计算的特点,加速回波模拟过程,大大提高了人体微波扫描成像***回波模拟的实时性,此外,由于使用GPU替代CPU进行回波模拟,从而可节省出CPU资源进行***控制、资源调度、逻辑分析等功能,另外,回波模拟时可根据需要更改仿真参数,使整个回波模拟仿真过程便捷易用,有助于一高***研制和调试效率,降低成本。
图5是本发明实施例提供的一种基于GPU的人体微波回波模拟***的示意性框图,该***用于运行本发明图3~图4所示实施例提供的方法。为了便于说明仅仅示出了与本实施例相关的部分。
参见图5所示,本实施例提供的一种基于GPU的人体微波回波模拟***,包括CPU主机端1和GPU设备端2,所述CPU主机端1包括仿真参数发送单元12、并行参数设置单元11以及内核函数启动单元13,所述GPU设备端2包括并行计算单元21和回波数据返回单元22,其中:
所述仿真参数发送单元12,用于将仿真输入参数由CPU主机端1的内存传送至GPU设备端2的显存中;
所述并行参数设置单元11,用于在所述CPU主机端1设置运行在所述GPU设备端2的并行计算网络参数;
内核函数启动单元13,用于启动所述CPU主机端1内预置的用于人体微波回波模拟的内核函数;
所述并行计算单元21,用于在所述内核函数控制下根据所述并行计算网络参数以多线程的方式在所述GPU设备端2的多个处理核中并行运行计算得出人体微波的模拟回波;
所述回波数据返回单元22,用于将所述GPU设备端2计算得出的人体微波的模拟回波回传至所述GPU主机端。
可选的,所述仿真输入参数包括:人体扫描成像***的扫描参数、天线相位中心位置参数以及人体模型分布式目标参数。
可选的,所述并行计算网络参数包括所述GPU设备端2进行并行计算的线程块的维度以及每个线程块内线程的维度。
可选的,所述并行计算单元21具体用于:
根据当前GPU线程所在的线程块在所有线程块中所处的编号确 定该线程块内所有线程共同使用的天线相位中心位置;
根据当前GPU线程在其所属线程块中所处的编号确定当前GPU线程计算所用到的人体模型分布式目标散射点的位置坐标和散射系数大小;
计算所述天线相位中心位置和所述人体模型分布式目标散射点之间的距离,并根据所述距离计算出从所述人体模型分布式目标散射点到所述天线相位中心位置的时延;
确定所述人体分布式目标散射点到所述天线相位中心位置的时延在回波中所处的距离门单元位置;
将所述距离门单元内的回波信号加上来自所述人体分布式目标散射点的多普勒相位得到所述天线相位中心收到的所述距离门单元的回波信号数据;
返回所述根据当前GPU线程在其所属线程块中所处的编号确定当前GPU线程计算所用到的人体模型分布式目标散射点的位置坐标和散射系数大小的步骤,重复执行上述流程直到当前线程块内所有的线程均执行完毕,从而得到所述天线相位中心位置收到的所有距离门单元的回波信号数据;
将所述回波信号数据与所述人体扫描成像***的发射信号脉冲进行卷积计算,得到最终输出的采用该发射信号脉冲的回波信号数据;
在所述GPU设备端2执行的其它线程块内并行执行上述流程,最终得到对应所有天线相位中心位置的回波信号数据。
可选的,所述人体扫描成像***的发射信号脉冲为线性调频信号脉冲。
需要说明的是,本实施例提供的***中的各个单元由于本发明方法实施例基于同一构思,其带来的技术效果与本发明方法实施例相同,具体内容可参见本发明方法实施例中的叙述,此处不再赘述。
因此,可以看出本发明实施例提供的一种基于GPU的人体微波回波模拟***同样可以充分利用GPU可进行并行计算的特点,加速回波模拟过程,大大提高了人体微波扫描成像***回波模拟的实时性;此外,由于使用GPU替代CPU进行回波模拟,从而可节省出CPU资源进行***控制、资源调度、逻辑分析等功能,另外,回波模拟时可根据需要更改仿真参数,使整个回波模拟仿真过程便捷易用,有助于一高***研制和调试效率,降低成本。
以上所述仅为本发明的较佳实施例而已,并不用以限制本发明,凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等,均应包含在本发明的保护范围之内。

Claims (10)

  1. 一种基于GPU的人体微波回波模拟方法,其特征在于,包括:
    将仿真输入参数由中央处理单元CPU主机端的内存传送至图像处理单元GPU设备端的显存中;
    在所述CPU主机端设置运行在所述GPU设备端的并行计算网络参数;
    启动所述CPU主机端内预置的用于人体微波回波模拟的内核函数;
    所述内核函数根据所述并行计算网络参数以多线程的方式在所述GPU设备端的多个处理核中并行运行计算得出人体微波的模拟回波;
    所述GPU端将计算得出的人体微波的模拟回波回传至所述GPU主机端。
  2. 如权利要求1所述的方法,其特征在于,所述仿真输入参数包括:人体扫描成像***的扫描参数、天线相位中心位置参数以及人体模型分布式目标参数。
  3. 如权利要求2所述的方法,其特征在于,所述并行计算网络参数包括所述GPU设备端进行并行计算的线程块的维度以及每个线程块内线程的维度。
  4. 如权利要求3所述的方法,其特征在于,所述内核函数根据所述并行计算网络参数以多线程的方式在所述GPU设备端的多个处 理核中并行运行计算得出人体微波的模拟回波具体包括:
    根据当前GPU线程所在的线程块在所有线程块中所处的编号确定该线程块内所有线程共同使用的天线相位中心位置;
    根据当前GPU线程在其所属线程块中所处的编号确定当前GPU线程计算所用到的人体模型分布式目标散射点的位置坐标和散射系数大小;
    计算所述天线相位中心位置和所述人体模型分布式目标散射点之间的距离,并根据所述距离计算出从所述人体模型分布式目标散射点到所述天线相位中心位置的时延;
    确定所述人体分布式目标散射点到所述天线相位中心位置的时延在回波中所处的距离门单元位置;
    将所述距离门单元内的回波信号加上来自所述人体分布式目标散射点的多普勒相位得到所述天线相位中心收到的所述距离门单元的回波信号数据;
    返回所述根据当前GPU线程在其所属线程块中所处的编号确定当前GPU线程计算所用到的人体模型分布式目标散射点的位置坐标和散射系数大小的步骤,重复执行上述流程直到当前线程块内所有的线程均执行完毕,从而得到所述天线相位中心位置收到的所有距离门单元的回波信号数据;
    将所述回波信号数据与所述人体扫描成像***的发射信号脉冲进行卷积计算,得到最终输出的采用该发射信号脉冲的回波信号数据;
    在所述GPU设备端执行的其它线程块内并行执行上述流程,最终得到对应所有天线相位中心位置的回波信号数据。
  5. 如权利要求4所述的方法,其特征在于,所述人体扫描成像***的发射信号脉冲为线性调频信号脉冲。
  6. 一种基于GPU的人体微波回波模拟***,其特征在于,包括CPU主机端和GPU设备端,所述CPU主机端包括仿真参数发送单元、并行参数设置单元以及内核函数启动单元,所述GPU设备端包括并行计算单元和回波数据返回单元,其中:
    所述仿真参数发送单元,用于将仿真输入参数由CPU主机端的内存传送至GPU设备端的显存中;
    所述并行参数设置单元,用于在所述CPU主机端设置运行在所述GPU设备端的并行计算网络参数;
    内核函数启动单元,用于启动所述CPU主机端内预置的用于人体微波回波模拟的内核函数;
    所述并行计算单元,用于在所述内核函数控制下根据所述并行计算网络参数以多线程的方式在所述GPU设备端的多个处理核中并行运行计算得出人体微波的模拟回波;
    所述回波数据返回单元,用于将所述GPU设备端计算得出的人体微波的模拟回波回传至所述GPU主机端。
  7. 如权利要求6所述的***,其特征在于,所述仿真输入参数包括:人体扫描成像***的扫描参数、天线相位中心位置参数以及人体模型分布式目标参数。
  8. 如权利要求7所述的***,其特征在于,所述并行计算网络参数包括所述GPU设备端进行并行计算的线程块的维度以及每个线程块内线程的维度。
  9. 如权利要求8所述的***,其特征在于,所述并行计算单元具体用于:
    根据当前GPU线程所在的线程块在所有线程块中所处的编号确定该线程块内所有线程共同使用的天线相位中心位置;
    根据当前GPU线程在其所属线程块中所处的编号确定当前GPU线程计算所用到的人体模型分布式目标散射点的位置坐标和散射系数大小;
    计算所述天线相位中心位置和所述人体模型分布式目标散射点之间的距离,并根据所述距离计算出从所述人体模型分布式目标散射点到所述天线相位中心位置的时延;
    确定所述人体分布式目标散射点到所述天线相位中心位置的时延在回波中所处的距离门单元位置;
    将所述距离门单元内的回波信号加上来自所述人体分布式目标散射点的多普勒相位得到所述天线相位中心收到的所述距离门单元的回波信号数据;
    返回所述根据当前GPU线程在其所属线程块中所处的编号确定当前GPU线程计算所用到的人体模型分布式目标散射点的位置坐标和散射系数大小的步骤,重复执行上述流程直到当前线程块内所有的线程均执行完毕,从而得到所述天线相位中心位置收到的所有距离门 单元的回波信号数据;
    将所述回波信号数据与所述人体扫描成像***的发射信号脉冲进行卷积计算,得到最终输出的采用该发射信号脉冲的回波信号数据;
    在所述GPU设备端执行的其它线程块内并行执行上述流程,最终得到对应所有天线相位中心位置的回波信号数据。
  10. 如权利要求9所述的***,其特征在于,所述人体扫描成像***的发射信号脉冲为线性调频信号脉冲。
PCT/CN2017/096100 2016-10-09 2017-08-04 基于gpu的人体微波回波模拟方法及*** WO2018064917A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/340,332 US11550078B2 (en) 2016-10-09 2017-08-04 GPU-based human body microwave echo simulation method and system

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610882825.X 2016-10-09
CN201610882825.XA CN106646664B (zh) 2016-10-09 2016-10-09 基于gpu的人体微波回波模拟方法及***

Publications (1)

Publication Number Publication Date
WO2018064917A1 true WO2018064917A1 (zh) 2018-04-12

Family

ID=58854506

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/096100 WO2018064917A1 (zh) 2016-10-09 2017-08-04 基于gpu的人体微波回波模拟方法及***

Country Status (3)

Country Link
US (1) US11550078B2 (zh)
CN (1) CN106646664B (zh)
WO (1) WO2018064917A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112180458A (zh) * 2020-09-15 2021-01-05 北京子兆信息技术有限公司 基于mimo的成像仪天线布局方法、阵列及成像检测方法
CN114186468A (zh) * 2021-12-16 2022-03-15 西安交通大学 一种基于gpu加速的二次电子发射模拟方法及***及设备

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106646664B (zh) 2016-10-09 2018-10-23 华讯方舟科技有限公司 基于gpu的人体微波回波模拟方法及***
CN109388495A (zh) * 2018-11-01 2019-02-26 北京视甄智能科技有限公司 一种基于单gpu卡的图像并发处理方法、装置及***

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101199428A (zh) * 2007-04-18 2008-06-18 汕头超声仪器研究所 一种超声回波数据的处理装置及方法
US20100066595A1 (en) * 2008-09-18 2010-03-18 Lee Chul J Electromagnetic (em) solver using a shooting bouncing ray (sbr) technique
CN101859341A (zh) * 2009-04-13 2010-10-13 盛林 影像引导消融治疗手术规划装置
CN102788979A (zh) * 2012-07-20 2012-11-21 电子科技大学 一种基于后向投影InSAR成像配准的GPU实现方法
CN104688273A (zh) * 2015-03-16 2015-06-10 哈尔滨工业大学 基于cpu+gpu异构架构的超高速超声成像装置及方法
CN106646664A (zh) * 2016-10-09 2017-05-10 华讯方舟科技有限公司 基于gpu的人体微波回波模拟方法及***

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017105566A1 (en) * 2015-12-17 2017-06-22 Massachusetts Institute Of Technology Methods and systems for near-field microwave imaging

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101199428A (zh) * 2007-04-18 2008-06-18 汕头超声仪器研究所 一种超声回波数据的处理装置及方法
US20100066595A1 (en) * 2008-09-18 2010-03-18 Lee Chul J Electromagnetic (em) solver using a shooting bouncing ray (sbr) technique
CN101859341A (zh) * 2009-04-13 2010-10-13 盛林 影像引导消融治疗手术规划装置
CN102788979A (zh) * 2012-07-20 2012-11-21 电子科技大学 一种基于后向投影InSAR成像配准的GPU实现方法
CN104688273A (zh) * 2015-03-16 2015-06-10 哈尔滨工业大学 基于cpu+gpu异构架构的超高速超声成像装置及方法
CN106646664A (zh) * 2016-10-09 2017-05-10 华讯方舟科技有限公司 基于gpu的人体微波回波模拟方法及***

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112180458A (zh) * 2020-09-15 2021-01-05 北京子兆信息技术有限公司 基于mimo的成像仪天线布局方法、阵列及成像检测方法
CN112180458B (zh) * 2020-09-15 2024-03-15 北京子兆信息技术有限公司 基于mimo的毫米波人体安检成像仪天线的布局布阵方法
CN114186468A (zh) * 2021-12-16 2022-03-15 西安交通大学 一种基于gpu加速的二次电子发射模拟方法及***及设备
CN114186468B (zh) * 2021-12-16 2024-03-29 西安交通大学 一种基于gpu加速的二次电子发射模拟方法及***及设备

Also Published As

Publication number Publication date
US11550078B2 (en) 2023-01-10
CN106646664B (zh) 2018-10-23
CN106646664A (zh) 2017-05-10
US20200041691A1 (en) 2020-02-06

Similar Documents

Publication Publication Date Title
WO2018064917A1 (zh) 基于gpu的人体微波回波模拟方法及***
CN104865562B (zh) 基于混合模型的雷达非合作目标的识别方法
WO2018076884A1 (zh) 一种近距离微波成像方法及***
CN110596646B (zh) 一种基于mimo体制的提高雷达角度分辨率的布局及方法
CN109270525B (zh) 基于深度学习的穿墙雷达成像方法及***
CA3000134A1 (en) Method and system for simulating a radar image
US11385380B2 (en) System and method for debugging millimeter wave security inspection instrument
CN106707255B (zh) 一种相控阵雷达仿真***及方法
CN102854507B (zh) 一种基于gpu后向投影双站合成孔径雷达成像方法
KR102272411B1 (ko) 합성 개구면 레이더 영상의 시뮬레이션 영상 데이터베이스의 식별률을 높이기 위한 인공신경망 학습 방법 및 그 장치
CN112069713B (zh) 一种近场散射特性建模方法、电子设备及存储介质
CN112649806B (zh) 一种mimo雷达近场三维成像方法
Guo et al. Research on sensing matrix characteristics in microwave staring correlated imaging based on compressed sensing
Perdana et al. A survey of graphics processing unit (GPU) utilization for radar signal and data processing system
CN111896956A (zh) 基于fpga和dsp的实时微波关联成像装置及方法
CN109738890A (zh) 一种基于弹载双基sar距离多普勒图像生成地距图的方法
CN113075659B (zh) 一种平坦场景网格模型自适应分区预处理方法和***
Yang et al. An Embedded-GPU-Based Scheme for Real-Time Imaging Processing of Unmanned Aerial Vehicle Borne Video Synthetic Aperture Radar
CN114089304A (zh) 一种动态交会状态目标近场散射回波的求解方法
CN105759266B (zh) 一种基于图形处理单元的sar雷达回波获取方法
CN110456318A (zh) 一种目标回波模拟方法、***、存储介质及终端设备
CN114063176B (zh) 雷达成像方法、装置及计算机可读存储介质
US11422226B2 (en) Systems and methods for multipath beam nulling
ZHANG et al. Real-time Implementation of Joint Domain Localised Algorithm for High Frequency Surface Wave Radar using GPU
Chen et al. Airborne bistatic SAR echo simulator based on Multi-GPU platform

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17857798

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17857798

Country of ref document: EP

Kind code of ref document: A1