WO2020134338A1 - 毫米波太赫兹成像设备及物体识别分类方法 - Google Patents

毫米波太赫兹成像设备及物体识别分类方法 Download PDF

Info

Publication number
WO2020134338A1
WO2020134338A1 PCT/CN2019/110408 CN2019110408W WO2020134338A1 WO 2020134338 A1 WO2020134338 A1 WO 2020134338A1 CN 2019110408 W CN2019110408 W CN 2019110408W WO 2020134338 A1 WO2020134338 A1 WO 2020134338A1
Authority
WO
WIPO (PCT)
Prior art keywords
resolution
array
polarization
polarized
millimeter wave
Prior art date
Application number
PCT/CN2019/110408
Other languages
English (en)
French (fr)
Inventor
赵自然
游�燕
张丽
王迎新
乔灵博
Original Assignee
同方威视技术股份有限公司
清华大学
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 同方威视技术股份有限公司, 清华大学 filed Critical 同方威视技术股份有限公司
Publication of WO2020134338A1 publication Critical patent/WO2020134338A1/zh

Links

Images

Classifications

    • 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
    • G01V8/20Detecting, e.g. by using light barriers using multiple transmitters or receivers

Definitions

  • the present disclosure relates to the technical field of security inspection, in particular to a millimeter wave terahertz imaging device, and a method for detecting objects to perform object recognition and classification using the millimeter wave terahertz imaging device.
  • Existing passive millimeter-wave terahertz imaging is similar to optical imaging. It uses a two-dimensional array to form a gaze at the target field of view, without scanning, and real-time imaging can be achieved.
  • the detector or radiometer or detector
  • the detector on each array element corresponds to a pixel, and the array element forms an array.
  • the detector or radiometer or detector performs Direct detection or indirect detection.
  • the resolution (object direction) of the passive human body security transposition is generally only 2-3cm. This resolution is not perfect for object classification and object recognition by size and shape.
  • the purpose of the present disclosure is to solve at least one aspect of the above technical problems, to provide a millimeter wave terahertz imaging device and an object recognition and classification method using the same.
  • the millimeter wave terahertz imaging device can identify objects to classify the objects on the basis of not generating harmful radiation to the human body, and the size of the identified objects can reach a millimeter-level structure.
  • a millimeter wave terahertz imaging apparatus for performing a security inspection on a subject, which includes a focusing lens, a detector, and a graphics processing device, wherein the focusing lens is provided on the subject Between the object and the detector, and is configured to focus the millimeter wave terahertz wave spontaneously radiated or reflected by the detected object on the detector;
  • the detector includes an antenna array and a detector array , Wherein the antenna array is arranged on the side of the detector array facing the focusing lens and is arranged as an antenna port of the detector array, the detector array is arranged on the focal plane of the focusing lens, and is Configured to convert the millimeter wave terahertz wave received by the antenna array into a polarized image of the object to be inspected; and the graphics processing device is disposed on a side of the detector array away from the antenna array, and is configured To process the polarized image to identify and classify the object under inspection.
  • the antenna array includes a plurality of receiving antennas, each of the plurality of receiving antennas is linearly polarized or circularly polarized.
  • the detector array includes a plurality of wave sensing units, the number of the plurality of wave sensing units is the same as the number of the plurality of receiving antennas, and each wave sensing on the detector array The position of the unit corresponds to the position of each receiving antenna on the antenna array.
  • the antenna array is a one-dimensional array
  • the detector array is a one-dimensional array
  • the one-dimensional antenna array includes a plurality of macro pixel units arranged linearly, wherein each macro The pixel unit is an N*1 antenna array, where N is a positive integer and N ⁇ 3, and each macro pixel unit includes at least N-1 different polarization angles.
  • the antenna array is a two-dimensional array
  • the detector array is a two-dimensional array
  • the two-dimensional antenna array includes a plurality of macro pixel units arranged on a two-dimensional plane
  • the N receiving antennas of each macro pixel unit include at least one of the following: N linearly polarizing receiving antennas; N-1 linearly polarizing receiving antennas and a circle Polarized receiving antenna.
  • the polarization angles of the N linearly polarized receiving antennas are Deg1, Deg2, Deg3, ... DegN, respectively
  • i is a positive integer less than or equal to N.
  • the polarization angles of the N-1 linearly polarized receiving antennas are Deg1, Deg2, Deg3, ... DegN-1, respectively
  • i is a positive integer less than or equal to N-1;
  • the circular polarization includes at least one of left-hand circular polarization and right-hand circular polarization.
  • the millimeter wave terahertz imaging device further includes a millimeter wave terahertz radiation source, which is used to radiate the millimeter wave terahertz wave to the subject.
  • the antenna array is a one-dimensional array
  • the detector array is a one-dimensional array
  • the millimeter wave terahertz imaging device further includes a focus lens disposed between the object to be inspected and the focusing lens Rotatable scanning mirror in the optical path.
  • the rotatable scanning mirror can be rotated to image the first part on the object under inspection at the first rotation angle at the first specific sensible wave of the one-dimensional detector array On the unit, and at a second rotation angle different from the first rotation angle, the second part different from the first part on the detected object is imaged on the second part of the one-dimensional detector array, which is different from the first wave sensing unit On the wave unit.
  • a method for object recognition classification using the millimeter wave terahertz imaging device including:
  • the millimeter wave terahertz wave spontaneously radiated or reflected by the object to be inspected is received by the antenna array and focused on the detector array;
  • the millimeter wave terahertz wave received by the antenna array is converted into a polarized image of the detected object through the detector array;
  • both the receiving antenna array and the detector array are two-dimensional arrays.
  • the antenna array includes a plurality of macro pixel units, each macro pixel unit includes N receiving antennas, and the N receiving antennas have at least N-1 polarization angles, the detection The array includes the number of sensing units equal to the number and position of the receiving antenna, where N is a positive integer greater than or equal to 4,
  • the step of using the image processing device to process the polarized image to obtain a high-resolution polarized image includes:
  • N low-resolution polarized images are extracted from the pixels corresponding to the multiple sensing units, and each low-resolution polarized image has a polarized angle and includes All pixels with a same polarization angle;
  • the average value is calculated from the estimated non-polarization intensity data, and the average value is used as the non-polarization intensity value of each polarization unit with the corresponding polarization angle, for the entire array range Perform the same process in the internal, that is, N low-resolution non-polarized images;
  • step S3 The N low-resolution images obtained in step S1 are interpolated to obtain N intermediate images with different polarization angles under the guidance of the low-resolution non-polarized images obtained in step S2, and then the N images Low-resolution non-polarized images are subtracted from the intermediate images to obtain N low-resolution poorly polarized images;
  • step S4 processing the N low-resolution polarization difference images obtained in step S3 using bilinear difference and upsampling processing methods to obtain N corresponding high-resolution polarization difference images;
  • step S5 Summing the N high-resolution polarization images obtained in step S4 and the high-resolution non-polarization images obtained in step S2, and finally obtaining N high-resolution polarization images.
  • the step of using the image processing device to process the polarized image to obtain a high-resolution polarized image further includes: S6: high resolution for the polarized information obtained in step S5 Super-resolution image processing algorithm for rate-polarized images to improve resolution.
  • a polarized image of the object to be inspected can be obtained.
  • a high-resolution image with polarization information can be obtained.
  • Polarization imaging technology can not only detect the structural information of the surface of the object, such as roughness and texture, but also detect the conductivity and refractive index of the surface of the object. This scheme is more effective than the existing passive terahertz imager To the surface intensity information of the object) provides more information, which is very useful for object classification and object recognition.
  • polarization information such as different surface textures, roughness, refractive index, conductivity, etc. of the material
  • suspicious objects of similar shapes and sizes can be identified, that is, identified and classified.
  • the size of the object recognizable by the millimeter wave terahertz imaging device according to the present disclosure can be reduced to the millimeter level.
  • FIG. 1 shows a passive millimeter wave terahertz imaging device according to an embodiment of the present disclosure.
  • FIG. 2 shows an active millimeter wave terahertz imaging device according to an embodiment of the present disclosure.
  • FIG. 3 shows an imaging principle diagram of a millimeter wave terahertz imaging device including a two-dimensional antenna array according to an embodiment of the present disclosure.
  • FIG. 4 shows an imaging principle diagram of a millimeter wave terahertz imaging device including a one-dimensional antenna array according to an embodiment of the present disclosure.
  • 5A and 5B show simplified schematic diagrams of a macro pixel unit of a two-dimensional antenna array according to an embodiment of the present disclosure.
  • 6A and 6B show simplified schematic diagrams of a macro pixel unit of a two-dimensional antenna array according to an embodiment of the present disclosure.
  • FIG. 7 shows a simplified schematic diagram of a macro pixel unit of a one-dimensional antenna array according to an embodiment of the present disclosure.
  • FIG. 8 shows a simplified schematic diagram of a macro pixel unit of a one-dimensional antenna array according to an embodiment of the present disclosure.
  • FIG 9 shows an image obtained by a detector array according to an embodiment of the present disclosure.
  • FIG. 10 shows four low-resolution polarized images extracted from the image obtained by the detector array according to one embodiment of the present disclosure.
  • FIG. 1 shows a passive millimeter wave terahertz imaging device according to the present disclosure.
  • a millimeter-wave terahertz imaging device is used to perform a security check on the subject 1, and it includes a focusing lens 3, a detector 4 and a graphics processing device 6.
  • a focusing lens 3 is provided between the subject 1 and the detector 4 and is configured to focus the millimeter wave terahertz wave 2 spontaneously radiated or reflected back by the subject on the detector 4.
  • the detector includes an antenna array 41 and a detector array 42 (as shown in FIGS.
  • the antenna array 41 is disposed on the side of the detector array 42 facing the focusing lens 3 and is disposed as the An antenna port of a detector array 42
  • the detector array 42 is disposed on the focal plane of the focusing lens 3, and is configured to convert the millimeter wave terahertz wave received by the antenna array into the pole of the subject 1 ⁇ image.
  • the graphics processing device 6 is disposed on a side of the detector array 42 away from the antenna array 41, and is configured to process the polarized image to identify and classify the detected object.
  • the detector of a typical millimeter wave terahertz imaging device is equipped with an antenna port.
  • the main purpose of the antenna port is to increase the received power and improve the reception efficiency.
  • the antenna array 41 is provided as an antenna port of the detector array 42 and communicates with the detector array 42 so that the detector 4 itself has the function of selecting the polarization direction.
  • the image processing device 6 includes an analog signal processor 61, a digital-to-analog converter (D/A converter) 62, a digital signal processor 63, and an image display 64.
  • the detector array 42 converts the incident millimeter wave terahertz wave into an electrical signal at each pixel, and sends it to the analog signal processor 61;
  • the analog signal processor 61 is used to receive the analog signal from the detector, and It is sent to the digital-to-analog converter 62;
  • the digital-to-analog converter 62 is used to receive the signal transmitted by the analog signal processor, perform digital-to-analog conversion on it, and then send it to the digital signal processor 63;
  • the digital signal processor 63 is used to Receive the information converted by the converter, and demosaicing it, and then display the image obtained after demosaicing to the image display 64. The method of demosaicing will be described in detail below.
  • the terahertz wave is an electromagnetic wave having a frequency in the range of 100 GHz to 10 THz (10000 GHz).
  • the terahertz wave is between microwave and visible light, and coincides with the millimeter wave in the long wave band and infrared rays in the short wave band.
  • the frequency band of the millimeter wave is 26.5 to 300 GHz, and the millimeter wave terahertz wave described in the present disclosure refers to electromagnetic waves with a frequency band between 30 GHz and 1000 GHz.
  • the millimeter wave terahertz wave is more suitable for safety inspection.
  • the active millimeter-wave terahertz imaging device further includes a millimeter-wave terahertz radiation source 5 for radiating a millimeter-wave terahertz wave to the subject 1 so that the subject 1 reflects toward the focusing lens 3 Millimeter wave terahertz wave.
  • the antenna array 41 includes a plurality of receiving antennas, and each of the plurality of receiving antennas is linearly polarized or circularly polarized.
  • the type, structure, and placement of the antenna determine the polarization (polarization direction) of the antenna.
  • Common antennas in the art include horn antennas, patch antennas, helical antennas, and so on.
  • the polarization angle of the horn antenna can be changed by setting the installation direction of the horn antenna and the patch antenna relative to the horizontal direction.
  • horn antennas and patch antennas can achieve different linear polarization directions through different non-horizontal placement methods.
  • the horn antenna may have a rectangular or circular waveguide opening. By setting the waveguide port as a circular waveguide port, the polarization mode of the horn antenna can be converted into circular polarization.
  • horn antennas and patch antennas can achieve circular polarization by adding dielectric wave plates or structural design.
  • the size of each side of the waveguide port of the horn antenna is preferably 0.1 mm to 10 mm to adapt to the wave sensing unit of the millimeter wave terahertz wave detector of different sizes.
  • the detector array 42 includes a plurality of wave sensing units, and the number of the plurality of wave sensing units is the same as the number of the plurality of receiving antennas, and each wave sensing on the detector array The position of the unit corresponds to the position of each receiving antenna on the antenna array.
  • the pixel pitch of the antenna array matches the pixel pitch of the detector array; crosstalk between adjacent wave sensing units of the detector array (mixed polarization information between adjacent pixels) As small as possible.
  • the antenna array may be a two-dimensional array or a one-dimensional array.
  • the two-dimensional antenna array 41 when the antenna array is a two-dimensional array, the two-dimensional antenna array 41 includes a plurality of macro pixel units arranged on a two-dimensional plane, where each macro pixel unit is M 1 *M 2
  • the antenna array of, where M 1 and M 2 are positive integers, and M 1 and M 2 ⁇ 2, and each macro pixel unit includes at least N-1 different polarization angles, where N M 1 *M 2 .
  • M is equal to 2
  • each macro pixel unit is a 2*2 antenna array.
  • the one-dimensional antenna array 41 when the antenna array is a one-dimensional array, the one-dimensional antenna array 41 includes a plurality of macro pixel units arranged linearly, wherein each macro pixel unit is an N*1 antenna array, where N It is a positive integer, and N ⁇ 3, and each macro pixel unit includes at least N-1 different polarization angles.
  • N is equal to 3
  • a macro pixel unit is an antenna array of 3*1.
  • different antenna types, antenna structures, and different antenna placement methods are selected according to the required size and polarization direction of the macro pixel unit.
  • a terahertz wave detector with a center frequency of 94 GHz is used, wherein the size of the antenna array is 120 ⁇ 160, and the size of the horn mouth of the horn antenna is (rectangular waveguide mouth) 5.5 mm ⁇ 4 cm, And the resolution of the detector array is 120 ⁇ 160, and the pixel size is 5mm ⁇ 5mm.
  • FIG. 3 shows an imaging principle diagram of a millimeter wave terahertz imaging device including a two-dimensional antenna array according to an embodiment of the present disclosure.
  • FIG. 4 shows an imaging principle diagram of a millimeter wave terahertz imaging device including a one-dimensional antenna array according to an embodiment of the present disclosure.
  • the antenna array is a two-dimensional antenna array
  • the detector array is a two-dimensional detector array.
  • the antenna array and the detector array are separated by a certain distance.
  • each antenna serves as an antenna port of each sensing unit.
  • the two-dimensional antenna array serves as the antenna port of the detector array.
  • the millimeter wave terahertz wave from the person's head is imaged on the one or more first sensory waves of the two-dimensional detector array after passing through the focusing lens and being received by the two-dimensional antenna array unit.
  • the millimeter wave terahertz wave from the human chest is imaged on one or more second wave sensing units of the two-dimensional detector array after passing through the lens and being received by the two-dimensional antenna array.
  • the location of a wave sensing unit is different. Therefore, the entire two-dimensional detector array can detect and image the millimeter wave terahertz waves from a plurality of different positions of the detected object at the same time.
  • the antenna array is a one-dimensional antenna array
  • the detector array is a one-dimensional detector array.
  • the one-dimensional antenna array serves as the antenna port of the detector array.
  • the millimeter wave terahertz imaging device further includes a rotatable scanning mirror 7 provided in the optical path between the subject 1 and the focus lens 3. The rotatable scanning mirror 7 can be rotated to image a specific part on the object under inspection at a specific rotation angle on a specific wave sensing unit of the one-dimensional detector array.
  • the millimeter wave terahertz imaging device images the head of the examinee on the first wave sensing unit of the one-dimensional detector array.
  • the millimeter wave terahertz imaging device images the chest of the examinee and other parts of the one-dimensional detector array and the first wave sensing unit On a different second wave sensor unit.
  • the rotatable scanning mirror 7 is repeatedly rotated until the entire scanning of the object to be inspected is achieved, and each part is imaged on the one-dimensional detector array.
  • the identification and classification of detected objects is the main goal of millimeter wave detection research.
  • the electromagnetic waves radiated by the object under inspection have polarization characteristics, so more information about the object under inspection can be obtained through the polarization information in the radiation signal of the object under inspection.
  • the present disclosure proposes to control the polarization of different detectors by the type of antenna or the placement of the antenna, that is to say, in a macro pixel unit of the antenna array or detector array, different wave sensing units receive different polarization states wave.
  • This sub-focal plane polarization imaging technology can infinitely add a receiving antenna array, and has a simple structure.
  • the polarization information of the detected polarization image can be used for object classification and object recognition.
  • the polarization imaging technology can not only detect the structural information of the surface of the object, such as roughness and texture, but also detect the conductivity and refractive index of the surface of the object.
  • This solution is more effective than the existing passive terahertz imager ( Only the surface intensity information of the object can be detected) provides more information, which is very useful for object classification and object recognition.
  • a common passive millimeter wave terahertz security detector is used to detect suspicious objects carried by the human body, such as mobile phones, banknotes, knives, and pistols.
  • the grayscale image shows that the human body is white, and Suspicious objects are black blocks.
  • the obtained polarization information (different surface textures, roughness, refractive index, conductivity, etc. of materials) is used to distinguish suspicious objects of similar shapes and sizes.
  • super-resolution imaging can be achieved through the super-resolution polarization imaging reconstruction algorithm, the resolution is at least 4 times higher than the existing imaging image mode (polarization information cannot be obtained), and the resolution can reach the millimeter level. It is very effective for identifying suspicious objects in millimeter-level structures.
  • the macro pixel unit includes N linearly polarized receiving antennas, and their polarization angles are Deg1, Deg2, Deg3, ... DegN, respectively.
  • i is a positive integer less than or equal to N.
  • the arrangement of macro pixels in one macro pixel unit is linear polarization of 0°, 45°, 90°, and -45°.
  • the macro pixel arrangement of one macro pixel unit is linear polarization at 30°, 75°, 120°, and -15°.
  • the macro pixel unit includes N-1 linearly polarized receiving antennas and one circularly polarized receiving antenna.
  • the circular polarization may be left-handed circular polarization or right-handed circular polarization, N-1
  • the linear polarization angles are Deg1, Deg2, Deg3, ... Deg(N-1), where or i is a positive integer less than or equal to N-1. As shown in FIG.
  • the polarization angles of the four receiving antennas are 0° linear polarization, 60° linear polarization, 120° linear polarization and circular polarization Change.
  • the polarization angles of the four receiving antennas are 0° linear polarization, 45° linear polarization, 90° linear polarization and circular polarization Change.
  • the macro pixel unit includes N linearly polarized receiving antennas whose polarization angles are Deg1, Deg2, Deg3, ... DegN, respectively
  • N the number of receiving antennas of a macro pixel unit
  • the arrangement of macro pixels in a macro pixel unit is linear polarization of 0°, 60°, and 120°.
  • FIG. 8 shows a simplified schematic diagram of a macro pixel unit of a one-dimensional antenna array according to an embodiment of the present disclosure.
  • the macro pixel unit includes N-1 linearly polarized receiving antennas and one circularly polarized receiving antenna.
  • the circular polarization may be left-handed circular polarization or right-handed circular polarization, N-1
  • a method for object recognition classification using the millimeter wave terahertz imaging device described above includes the following steps: through the focusing lens, the millimeter wave terahertz wave spontaneously radiated or reflected by the object to be detected is received by the antenna array and focused on the detector array; through the detector array , Convert the millimeter wave terahertz wave received by the antenna array into a polarized image of the subject (for example, the polarized image shown in FIG. 9); use the image processing device to process the polarized image to obtain High-resolution polarized image; based on the obtained high-resolution polarized image, automatic recognition algorithm is used for object recognition classification.
  • the receiving antenna array and the detector array are both two-dimensional arrays. It can be understood that the receiving antenna array and the detector array may also adopt a one-dimensional array.
  • the millimeter wave terahertz imaging device further includes a rotatable scanning mirror disposed between the focusing lens and the object to be inspected. The function and working mode of the rotatable scanning mirror have been described in detail above, and will not be repeated here.
  • the polarized image will be processed by an image processing device to obtain a high-resolution polarized image
  • the antenna array includes multiple macro pixel units, each macro pixel unit includes N receiving antennas, and the N receiving antennas have at least N-1 polarization angles
  • the detector array includes Wave sensing units with equal number of antennas and corresponding positions (N is a positive integer greater than or equal to 4).
  • the image processing device is used to process the polarized image to obtain a high-resolution polarized image, thereby completing the demosaic processing of the original image.
  • the processing procedure includes the following five steps.
  • step S1 in the polarized image obtained by the detector array (for example, as shown in FIG. 9), N low-resolution polarized images are extracted from the pixels corresponding to the multiple sensing units (for example, as shown in FIG. 10) (4 images shown in (a), (b), (c), and (d)), each low-resolution polarized image has a polarization angle and includes all pixels with a same polarization angle point. For example, as shown in FIG.
  • the adopted detector array includes 16 sensing units, so the resolution of the obtained polarized image is 4*4.
  • the antenna array corresponding to the detector array includes 16 receiving antennas, so the antenna array size is 4*4, and the polarization angles of the macro pixel units of the antenna array are -45° linear polarization, 0° linear polarization, 45 °Linear polarization and 90°line polarization, and the resolution of the macro pixel unit is 2*2. Therefore, as shown in FIG.
  • the resolutions of the four low-resolution polarized images obtained through step S1 are all 2*2, and (a) the polarization angle of the image is 0°, (b) the polarization of the image The angle is 45°, (c) the polarization angle of the image is 90° and (d) the polarization angle of the image is -45°.
  • the non-polarization intensity data of the pixel at the polarization angle position in the polarization array is estimated to obtain a high-resolution non-polarization image whose size is equal to the size of the antenna array.
  • the average value is calculated from the estimated non-polarization intensity data, and the average value is used as the non-polarization intensity value of each polarization unit with the corresponding polarization angle, for the entire array range.
  • N low-resolution non-polarized images are obtained.
  • the resolution of the high-resolution non-polarized image is 4*4
  • the number of low-resolution non-polarized images is 4 and the resolution is 2*2.
  • step S3 the N low-resolution images obtained in step S1 are guided by the low-resolution non-polarized images obtained in step S2, and N intermediate images with different polarization angles are obtained by interpolation, and then N intermediate images are subtracted from the low-resolution non-polarized images, respectively, to obtain N low-resolution poorly polarized images; as shown in FIG. 10, in the illustrated embodiment, four resolutions are obtained with a resolution of 2* 2 low resolution poorly polarized images
  • step S4 the N low-resolution polarization difference images obtained in step S3 are processed using bilinear difference and upsampling processing methods to obtain N corresponding high-resolution polarization difference images. As shown in FIG. 10, in the illustrated embodiment, four high-resolution polarized image with a resolution of 4*4 are obtained.
  • step S5 the N high-resolution polarized difference images obtained in step S4 and the high-resolution non-polarized image obtained in step S2 are summed to finally obtain N high-resolution polarized images.
  • N high-resolution polarized images As shown in FIG. 10, in the illustrated embodiment, four high-resolution polarized images are obtained.
  • a super-resolution image processing algorithm may be performed on the high-resolution polarized image with polarization information to increase the resolution.
  • Super-resolution imaging can be achieved through the super-resolution polarization imaging reconstruction algorithm, the resolution is at least 4 times higher than the existing imaging image mode (polarization information cannot be obtained), and the resolution can reach the millimeter level. This is very effective for identifying suspicious objects in millimeter-level structures.

Landscapes

  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Geophysics (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)

Abstract

一种用于对被检对象(1)进行安全检查的毫米波太赫兹成像设备,包括聚焦透镜(3),检测器(4)和图形处理装置(6),其中聚焦透镜(3)设置在被检对象(1)和检测器(4)之间,且被构造为将被检对象(1)自发辐射或反射回来的毫米波太赫兹波聚焦在检测器(4)上;检测器(4)包括天线阵列(41)和探测器阵列(42),其中天线阵列(41)设置在探测器阵列(42)的朝向聚焦透镜(3)的一侧且设置为探测器阵列(41)的天线端口,探测器阵列(42)设置在聚焦透镜(3)的焦平面上,且被构造为将天线阵列(41)接收的毫米波太赫兹波转化为被检对象(1)的极化图像;图形处理装置(6)设置于探测器阵列(42)的远离天线阵列(41)的一侧,且被构造为处理极化图像以对被检对象(1)进行识别分类。

Description

毫米波太赫兹成像设备及物体识别分类方法
本申请要求于2018年12月29日递交中国专利局的、申请号为201811654183.3的中国专利申请的权益,这些申请的全部公开内容以引用方式并入本文。
技术领域
本公开涉及安检技术领域,特别是涉及一种毫米波太赫兹成像设备,以及利用上述毫米波太赫兹成像设备对物体进行检测以进行物体识别分类的方法。
背景技术
在现有的被动式毫米波太赫兹成像类似于光学摄像,利用一个二维阵面对目标视场形成凝视,不需要扫描,可实现实时成像。在该二维阵面上,每个阵元上的探测器(或者辐射计,或者检波器)对应一个像素,由阵列形式的阵元构成一个阵面,该探测器或辐射计或检波器进行直接检测或者进行间接检测。
考虑到毫米波太赫兹探测器的成本,完全采取二维焦平面直接成像方式将导致整个系统成本十分昂贵。所以,在实际应用中为了同时兼顾系统成本和成像速率的要求,针对二维成像,当前的主流系统均采用一定数量的辐射计加上机械扫描的方式实现对整个视场的扫描覆盖,通过牺牲成像时间来减低少对探测器数目的需求,从而降低整个系统的成本。
现有的基于焦平面成像的被动式毫米波太赫兹成像安检装置无论是采用辐射计的直接探测还是外差法的间接探测,都只能通过可疑物(如手机、钞票、刀具、手枪等)与人体之间的温度差显示可疑物的图像形状,进而确定人体是否携带可疑物,而无法对可疑物进行物体识别。通常人体体表温度比可疑物高,在成像灰度图像上显示人体是白色,而可疑物是黑色。通常,无论是机器识别还是人工识别,均无法将类似形状和大小的皮带扣、手机、金属块、介质块和纸币等进行物体识别。
另外目前被动式人体安检转置的分辨率(物方向)一般只有2-3cm,这个分辨率对于通过大小和形状进行物体分类和物体识别是不完善的。
发明内容
本公开的目的在于解决上述技术问题中的至少一个方面,提供一种毫米波太赫兹成像设备及其利用该设备进行的物体识别和分类方法。通过该毫米波太赫兹成像设备 能够在不对人体产生有害辐射的基础上,识别出物体以对物体进行分类,且识别到的物体的大小能够达到毫米级的结构。
在根据本公开的一个方面中,提供一种用于对被检对象进行安全检查的毫米波太赫兹成像设备,其包括聚焦透镜,检测器和图形处理装置,其中所述聚焦透镜设置在被检对象和所述检测器之间,且被构造为将被检对象自发辐射或反射回来的毫米波太赫兹波聚焦在所述检测器上;检测器,所述检测器包括天线阵列和探测器阵列,其中天线阵列设置在所述探测器阵列的朝向所述聚焦透镜的一侧且设置为所述探测器阵列的天线端口,所述探测器阵列设置在所述聚焦透镜的焦平面上,且被构造为将所述天线阵列接收的毫米波太赫兹波转化为被检对象的极化图像;以及所述图形处理装置设置于所述探测器阵列的远离所述天线阵列的一侧,且被构造为处理所述极化图像以对被检对象进行识别分类。
根据本公开的一个示例性实施例,所述天线阵列包括多个接收天线,所述多个接收天线中的每个接收天线都被线极化或被圆极化。
根据本公开的另一个示例性实施例,所述探测器阵列包括多个感波单元,多个感波单元的数量与多个接收天线的数量相同,所述探测器阵列上的每个感波单元的位置与所述天线阵列上的每个接收天线的位置相对应。
根据本公开的另一个示例性实施例,所述天线阵列为一维阵列,所述探测器阵列为一维阵列,所述一维天线阵列包括线性排列的多个宏像素单元,其中每个宏像素单元为N*1的天线阵列,其中N为正整数,且N≥3,且每个宏像素单元包括至少N-1个不同的极化角度。
根据本公开的另一个示例性实施例,所述天线阵列为二维阵列,所述探测器阵列为二维阵列,所述二维天线阵列包括在二维平面上排列的多个宏像素单元,其中每个宏像素单元为M 1*M 2的天线阵列,其中M 1,M 2为正整数,且M 1,M 2≥2,且每个宏像素单元包括至少N-1个不同的极化角度,其中N=M 1*M 2
根据本公开的另一个示例性实施例,每个宏像素单元的N个接收天线包括如下方式中的至少一种:N个线极化接收天线;N-1个线极化接收天线和一个圆极化接收天线。
根据本公开的另一个示例性实施例,N个线极化接收天线的极化角度分别为Deg1、Deg2、Deg3、…DegN,其中
Figure PCTCN2019110408-appb-000001
其中i为小于等于N的正整数。
根据本公开的另一个示例性实施例,N-1个线极化接收天线的极化角度分别为Deg1、Deg2、Deg3、…DegN-1,其中
Figure PCTCN2019110408-appb-000002
Figure PCTCN2019110408-appb-000003
其中i为小于等于N-1的正整数;
其中,圆极化包括左旋圆极化和右旋圆极化中的至少一种。
根据本公开的另一个示例性实施例,毫米波太赫兹成像设备还包括毫米波太赫兹辐射源,其用于向被检对象辐射毫米波太赫兹波。
根据本公开的另一个示例性实施例,所述天线阵列为一维阵列,所述探测器阵列为一维阵列,所述毫米波太赫兹成像设备还包括设置在被检对象和聚焦透镜之间的光路中的可旋转扫描反射镜。
根据本公开的另一个示例性实施例,所述可旋转扫描反射镜能够旋转,从而在第一旋转角度下将被检对象上的第一部位成像在一维探测器阵列的第一特定感波单元上,且在与第一旋转角度不同的第二旋转角度下将被检对象上的与第一部位不同的第二部位成像在一维探测器阵列的与第一感波单元不同的第二感波单元上。
根据本公开的另一方面,提供一种使用根据上述的毫米波太赫兹成像设备进行物体识别分类的方法,包括:
通过所述聚焦透镜,使得被检对象自发辐射或反射回来的毫米波太赫兹波被所述天线阵列接收且聚焦在所述探测器阵列上;
通过所述探测器阵列,将所述天线阵列接收的毫米波太赫兹波转化为被检对象的极化图像;
利用所述图像处理装置处理所述极化图像以获得高分辨率极化图像;
基于获得的高分辨率极化图像,利用自动识别算法进行物体识别分类。
根据本公开的一个示例性实施例,所述接收天线阵列和所述探测器阵列均是二维阵列。
根据本公开的另一个示例性实施例,所述天线阵列包括多个宏像素单元,每个宏像素单元包括N个接收天线,N个接收天线具有至少N-1个极化角度,所述探测器阵列包括与接收天线数量相等且位置对应的感波单元,其中N为大于等于4的正整数,
其中,利用所述图像处理装置处理所述极化图像以获得高分辨率极化图像的步骤 包括:
S1:在探测器阵列获得的极化图像中,从多个感波单元所对应的像素点中提取N幅低分辨率极化图像,每幅低分辨率极化图像具有一个极化角度且包括具有一个相同极化角度的所有像素点;
S2:估算出极化阵列中极化角度位置处像素的无极化强度数据,得到一幅高分辨率无极化图像,高分辨率无极化图像的分辨率与天线阵列的大小相等,以及
在高分辨率无极化图像的各个极化单元中,通过估算出来的无极化强度数据求平均值,该平均值作为具有相应极化角度的各个极化单元的无极化强度数值,对整个阵列范围内进行同样的处理,即得到N幅低分辨率无极化图像;
S3:将经过步骤S1得到的N幅低分辨率图像在步骤S2处理得到的低分辨率无极化图像的指导下,通过插值得到N幅不同极化角度的中间图像,然后再在得到的N幅中间图像中分别减去低分辨率无极化图像,即得到N幅低分辨率极化差图像;
S4:采用双线性差值、上采样的处理方法对步骤S3得到的N幅低分辨率极化差图像进行处理,得到N幅相对应的高分辨率极化差图像;以及
S5:将步骤S4得到的N幅高分辨率极化差图像与步骤S2得到的高分辨率无极化图像进行求和,最终得到N幅高分辨率极化图像。
根据本公开的另一个实施例,利用所述图像处理装置处理所述极化图像以获得高分辨率极化图像的步骤还包括:S6:针对在步骤S5中得到的具有极化信息的高分辨率极化图像进行超分辨率图像处理算法提高分辨率。
在根据本公开的毫米波太赫兹成像设备和利用该设备进行的物体识别分类的方法中,通过设置一维或二维天线阵列,能够获得被检对象的极化图像。该极化图像通过图像处理设备处理之后,能够获得高分辨率的带有极化信息的图像。极化成像技术不仅能够探测到物体表面的结构信息,如粗糙度和纹理,还能够探测物体表面的电导率、折射率等信息,这种方案比现有的被动式太赫兹成像仪(只能探测到物体表面强度信息)提供了更多的信息,这些信息对物体分类和物体识别是非常有用的。通过获取的极化信息,例如材料不同表面纹理,粗糙度,折射率,电导率等,能够对类似形状和大小的可疑物进行辨别,也就是进行识别和分类。此外,根据本公开的毫米波太赫兹成像设备可识别的物体大小能够缩小到毫米级别。
附图说明
图1示出了根据本公开的一个实施例的被动式毫米波太赫兹成像设备。
图2示出了根据本公开的一个实施例的主动式毫米波太赫兹成像设备。
图3示出了根据本公开的一个实施例的包括二维天线阵列的毫米波太赫兹成像设备的成像原理图。
图4示出了根据本公开的一个实施例的包括一维天线阵列的毫米波太赫兹成像设备的成像原理图。
图5A和5B示出了根据本公开的一个实施例的二维天线阵列的宏像素单元的简化示意图。
图6A和6B示出了根据本公开的一个实施例的二维天线阵列的宏像素单元的简化示意图。
图7示出了根据本公开的一个实施例的一维天线阵列的宏像素单元的简化示意图。
图8示出了根据本公开的一个实施例的一维天线阵列的宏像素单元的简化示意图。
图9示出了根据本公开的一个实施例的探测器阵列获得的图像。
图10示出了根据本公开的一个实施例的从探测器阵列获得的图像中提取的4幅低分辨率极化图像。
具体实施方式
虽然将参照含有本公开的较佳实施例的附图充分描述本公开,但在此描述之前应了解本领域的普通技术人员可修改本文中所描述的公开,同时获得本公开的技术效果。因此,须了解以上的描述对本领域的普通技术人员而言为一广泛的揭示,且其内容不在于限制本公开所描述的示例性实施例。
另外,在下面的详细描述中,为便于解释,阐述了许多具体的细节以提供对本披露实施例的全面理解。然而明显地,一个或多个实施例在没有这些具体细节的情况下也可以被实施。在其他情况下,公知的结构和装置以图示的方式体现以简化附图。
图1示出了根据本公开的被动式毫米波太赫兹成像设备。如图1所示,毫米波太赫兹成像设备用于对被检对象1进行安全检查,其包括聚焦透镜3,检测器4和图形处理装置6。聚焦透镜3设置在被检对象1和所述检测器4之间,且被构造为将被检对象自发辐射或反射回来的毫米波太赫兹波2聚焦在所述检测器4上。所述检测器包括天线阵列41和探测器阵列42(如图3和4所示),其中天线阵列41设置在所述探测器阵列42的朝向所述聚焦透镜3的一侧且设置为所述探测器阵列42的天线端口,所述探测器阵列42设置在所述聚焦透镜3的焦平面上,且被构造为将所述天线阵列接 收的毫米波太赫兹波转化为被检对象1的极化图像。所述图形处理装置6设置于所述探测器阵列42的远离所述天线阵列41的一侧,且被构造为处理所述极化图像以对被检对象进行识别分类。
典型的毫米波太赫兹成像设备的检测器都带有天线端口,天线端口的主要用途是增大接收的功率和提升接收的效率。在本公开中,天线阵列41被设置作为检测器阵列42的天线端口且与探测器阵列42进行通信,使得检测器4本身具有选择极化方向的功能。
在根据本公开的一个实施例中,所述图像处理装置6包括模拟信号处理器61,数模转换器(D/A转换器)62,数字信号处理器63以及图像显示器64。探测器阵列42将入射的毫米波太赫兹波转化为每个像素点上的电信号,并发送至模拟信号处理器61;模拟信号处理器61用于接收探测器传来的模拟信号,并将其发送至数模转换器62;数模转换器62用于接收经模拟信号处理器传输来的信号,并对其进行数模转换再发送至数字信号处理器63;数字信号处理器63用于接收经转换器转换后的信息,并对其进行去马赛克处理,再将去马赛克处理后得到的图像显示至图像显示器64上,其中去马赛克处理的方法将在下文中进行详细说明。
在本公开中,太赫兹波是频率在100GHz至10THz(10000GHz)范围为的电磁波,太赫兹波介于微波和可见光之间,在长波段与毫米波重合,在短波段与红外线重合。毫米波的频段为26.5至300GHz,本公开所述的毫米波太赫兹波是指频段位于30GHz至1000GHz之间的电磁波。在毫米波太赫兹成像设备的技术领域中,由于人体辐射或反射的毫米波太赫兹波的能量是非常低的,因此毫米波太赫兹波用于安全检查是较合适的。
图2示出了根据本公开的主动式毫米波太赫兹成像设备。如图2所示,该主动式毫米波太赫兹成像设备还包括毫米波太赫兹辐射源5,其用于向被检对象1辐射毫米波太赫兹波,使得被检对象1向聚焦透镜3反射毫米波太赫兹波。
在根据本公开的一个实施例中,所述天线阵列41包括多个接收天线,所述多个接收天线中的每个接收天线被线极化或被圆极化。
天线的种类、结构和放置方式决定天线的极化(极化方向)。本领域常见的天线包括喇叭天线、贴片天线、螺旋天线等。通过设置喇叭天线、贴片天线相对于水平方向的设置方向既可以改变喇叭天线的极化角度。总之,喇叭天线、贴片天线等通过不同的非水平摆放方式即可实现不同的线极化方向。喇叭天线可以具有矩形或圆形波导口。 通过将波导口设置为圆形波导口,可将喇叭天线的极化方式转换为圆极化。此外,通过在矩形波导口中加入介质波片也可实现喇叭天线等的圆极化。总之,喇叭天线、贴片天线通过加入介质波片或结构设计即可实现圆极化。在本公开中,喇叭天线的波导口的各个边的大小优选地为0.1mm至10mm,以适应不同大小的毫米波太赫兹波探测器的感波单元。
在根据本公开的一个实施例中,所述探测器阵列42包括多个感波单元,多个感波单元的数量与多个接收天线的数量相同,所述探测器阵列上的每个感波单元的位置与所述天线阵列上的每个接收天线的位置相对应。在根据本公开的毫米波太赫兹成像设备中,天线阵列的像素间距与探测器阵列像素间距相匹配;探测器阵列的相邻感波单元间的串扰(相邻像素间的混合极化信息)尽可能小。
在根据本公开的一个实施例中,天线阵列可以是二维阵列或一维阵列。
在根据本公开的一个实施例中,在天线阵列为二维阵列时,二维天线阵列41包括在二维平面上排列的多个宏像素单元,其中每个宏像素单元为M 1*M 2的天线阵列,其中M 1,M 2为正整数,且M 1,M 2≥2,且每个宏像素单元包括至少N-1个不同的极化角度,其中N=M 1*M 2。在一个具体的实施例中,M等于2,每个宏像素单元为2*2的天线阵列。
在根据本公开的一个实施例中,在天线阵列为一维阵列时,一维天线阵列41包括线性排列的多个宏像素单元,其中每个宏像素单元为N*1的天线阵列,其中N为正整数,且N≥3,且每个宏像素单元包括至少N-1个不同的极化角度。在一个具体的实施例中,N等于3,一个宏像素单元为3*1的天线阵列。
在本公开的毫米波太赫兹成像设备中,根据所需的宏像素单元的大小以及极化方向,来选择不同的天线种类、天线结构以及不同的天线放置方法。
在根据本公开的一个具体实施例中,使用中心频率为94GHz的太赫兹波探测器,其中天线阵列大小为120×160,喇叭天线的喇叭口的尺寸为(矩形波导口)5.5mm×4cm,并且探测器阵列分辨率为120×160,像元大小为5mm×5mm。
图3示出了根据本公开的一个实施例的包括二维天线阵列的毫米波太赫兹成像设备的成像原理图。图4示出了根据本公开的一个实施例的包括一维天线阵列的毫米波太赫兹成像设备的成像原理图。
在根据本公开的一个实施例中,如图3所示,所述天线阵列为二维天线阵列,所述探测器阵列为二维探测器阵列,在该示意图中,为了更清楚地示出天线阵列和探测 器阵列的结构示意图,天线阵列和探测器阵列相隔一定距离,然而在实际结构中,每个天线作为每个感测单元的天线端口。从而二维天线阵列作为探测器阵列的天线端口。例如,在被检对象为人的情况下,来自人的头部的毫米波太赫兹波在经过聚焦透镜和被二维天线阵列接收之后成像在二维探测器阵列的一个或多个第一感波单元。同时,来自人的胸部的毫米波太赫兹波在经过透镜和被二维天线阵列接收之后成像在二维探测器阵列的一个或多个第二感波单元上,该第二感波单元与第一感波单元的位置不同。因此,在整个二维探测器阵列能够在同一时刻检测来自被检对象的多个不同位置的毫米波太赫兹波并对其进行成像。在检测可疑对象时,能够对可疑对象上的多个位置辐射或反射的毫米波太赫兹波进行成像。
在根据本公开的一个实施例中,如图4所示,所述天线阵列为一维天线阵列,所述探测器阵列为一维探测器阵列。一维天线阵列作为探测器阵列的天线端口。在此情况下,该毫米波太赫兹成像设备还包括设置在被检对象1和聚焦透镜3之间的光路中的可旋转扫描反射镜7。该可旋转扫描反射镜7能够旋转以在一个特定旋转角度将被检对象上的特定部位成像在一维探测器阵列的特定感波单元上。例如,在可旋转扫描反射镜7处于第一旋转角度时,毫米波太赫兹成像设备将被检人的头部成像在一维探测器阵列的第一感波单元上。在可旋转扫描反射镜7处于与第一旋转角度不同的第二旋转角度时,毫米波太赫兹成像设备将被检人的胸部等其他部位成像在一维探测器阵列的与第一感波单元不同的第二感波单元上。重复旋转该可旋转扫描反射镜7,直到实现对被检对象的整体扫描,并将各个部位成像在一维探测器阵列上。通过设置该可旋转扫描反射镜7能够减少价格昂贵的探测器单元的数量,从而节省成本。
被检对象的识别和分类是毫米波探测研究的主要目标。被检对象辐射的电磁波具有极化特性,因此可通过被检对象辐射信号中的极化信息来获取更多的关于被检对象的信息。本公开提出了一种通过天线种类或者天线放置方式来控制不同检波器的极化度,也就是说天线阵列或探测器阵列的一个宏像素单元内,不同的感波单元接收不同极化状态的波。这种分焦平面极化成像技术能够无限额外添加接收天线阵列,结构简单。
利用根据本公开的毫米波太赫兹成像设备的优势主要体现在如下的两个方面。
第一方面,可以利用探测到的极化图像的极化信息进行物体分类和物体识别。这是因为极化成像技术不仅能够探测到物体表面的结构信息,如粗糙度和纹理,还能够探测物体表面的电导率、折射率等信息,这种方案比现有的被动式太赫兹成像仪(只 能探测到物体表面强度信息)提供了更多的信息,这些信息对物体分类和物体识别是非常有用的。例如采用常见的被动式毫米波太赫兹安检仪探测人体携带的可疑物,如手机、钞票、刀具和手枪等,由于人体体表温度比可疑物高,在成像灰度图像上显示人体是白色,而可疑物都是黑色块。通常,无论是机器识别还是人工识别,无法把类似形状和大小的皮带扣、手机、金属块、介质块和纸币进行区分。我们是无法通过黑块的形状来判别可疑物。但是,利用极化成像技术,用获取的极化信息(材料不同表面纹理,粗糙度,折射率,电导率等)对类似形状和大小的可疑物进行辨别。
另一方面,可通过超分辨率极化成像重构算法实现超分辨成像,分辨率比现有的成像图像模式(不能获得极化信息)的提高至少4倍,分辨率可以达到毫米级别,这对识别毫米级别结构的可疑物是非常有效的。
下文将详细说明一维天线阵列和二维天线阵列的极化方式。
图5A和5B示出了根据本公开的一个实施例的二维天线阵列的宏像素单元的简化示意图。在该实施例中,宏像素单元包括N个线极化接收天线,它们的极化角度分别是Deg1、Deg2、Deg3、…DegN,其中
Figure PCTCN2019110408-appb-000004
i为小于等于N的正整数。如图5A所示,每个宏像素单元的接收天线的数量N=4时,一个宏像素单元宏像素排列方式是0°、45°、90°和-45°的线极化。如图5B所示,一个宏像素单元宏像素排列方式是30°、75°、120°和-15°的线极化。
图6A和6B示出了根据本公开的一个实施例的二维天线阵列的宏像素单元的简化示意图。在该实施例中,宏像素单元包括N-1个线极化接收天线,与1个圆极化接收天线,圆极化可以是左旋圆极化也可以是右旋圆极化,N-1个线极化角度分别是Deg1,Deg2,Deg3,…Deg(N-1),其中
Figure PCTCN2019110408-appb-000005
Figure PCTCN2019110408-appb-000006
i为小于等于N-1的正整数。如图6A所示,每个宏像素单元的接收天线的数量N=4时,4个接收天线的极化角度是0°线极化、60°线极化、120°线极化和圆极化。如图6B所示,每个宏像素单元的接收天线的数量N=4时,4个接收天线的极化角度是0°线极化、45°线极化、90°线极化和圆极化。
图7示出了根据本公开的一个实施例的一维天线阵列的宏像素单元的简化示意图。在该实施例中,宏像素单元包括N个线极化接收天线,其极化角度分别是Deg1,Deg2,Deg3,…DegN,其中
Figure PCTCN2019110408-appb-000007
在一个实施例中,如图7所示,一个宏像 素单元的接收天线数量N=3时,一个宏像素单元宏像素排列方式是0°、60°和120°的线极化。
图8示出了根据本公开的一个实施例的一维天线阵列的宏像素单元的简化示意图。在该实施例中,宏像素单元包括N-1个线极化接收天线,与1个圆极化接收天线,圆极化可以是左旋圆极化也可以是右旋圆极化,N-1个线极化角度分别是Deg1,Deg2,Deg3,…Deg(N-1),其中
Figure PCTCN2019110408-appb-000008
在一个实施例中,如图8所示,一个宏像素单元的接收天线数量N=3时,一个宏像素单元宏像素排列方式是0°线极化、90°线极化和圆极化。
根据本公开的另一方面,还提供一种使用上述的毫米波太赫兹成像设备进行物体识别分类的方法。该方法包括如下的步骤:通过所述聚焦透镜,使得被检对象自发辐射或反射回来的毫米波太赫兹波被所述天线阵列接收且聚焦在所述探测器阵列上;通过所述探测器阵列,将所述天线阵列接收的毫米波太赫兹波转化为被检对象的极化图像(例如,如图9所示的极化图像);利用所述图像处理装置处理所述极化图像以获得高分辨率极化图像;基于获得的高分辨率极化图像,利用自动识别算法进行物体识别分类。
在根据本公开的一个实施例中,所述接收天线阵列和所述探测器阵列均是二维阵列。可以理解的是,所述接收天线阵列和所述探测器阵列也可采用一维阵列。在所述接收天线阵列和所述探测器阵列均是一维阵列的情况下,在所述毫米波太赫兹成像设备还包括一个设置在聚焦透镜和被检对象之间的可旋转扫描反射镜。该可旋转扫描反射镜的功能以及工作方式在上文已经详细说明,在此不再赘述。
在根据本公开的一个实施例中,在所述接收天线阵列和所述探测器阵列均是二维阵列的情况下,将对图像处理装置处理所述极化图像以得到高分辨率极化图像的方法进行详细说明,也就是图像去马赛克处理的方法。在该实施例中,所述天线阵列包括多个宏像素单元,每个宏像素单元包括N个接收天线,N个接收天线具有至少N-1个极化角度,所述探测器阵列包括与接收天线数量相等且位置对应的感波单元(N为大于等于4的正整数)。
在该实施例中,利用所述图像处理装置处理所述极化图像以获得高分辨率极化图像,从而完成原始图像的去马赛克处理,该处理过程包括如下的5个步骤。在步骤S1中,在探测器阵列获得的极化图像(例如如图9所示)中,从多个感波单元所对应的 像素点中提取N幅低分辨率极化图像(例如如图10所示的(a),(b),(c),(d)所示的4幅图像),每幅低分辨率极化图像具有一个极化角度且包括具有一个相同极化角度的所有像素点。例如,如图9所示,所采用的探测器阵列包括16个感测单元,因此获得的极化图像的分辨率为4*4。该探测器阵列对应的天线阵列包括16个接收天线,因此天线阵列大小为4*4,且天线阵列的宏像素单元的极化角度分别为-45°线极化,0°线极化,45°线极化和90°线极化,且宏像素单元的分辨率为2*2。因此,如图10所示,通过步骤S1获得的4幅低分辨率极化图像的分辨率均为2*2,且(a)图像的极化角度为0°,(b)图像的极化角度为45°,(c)图像的极化角度为90°和(d)图像的极化角度为-45°。
在步骤S2处,估算出极化阵列中极化角度位置处像素的无极化强度数据,得到一幅高分辨率无极化图像,高分辨率无极化图像的大小与天线阵列的大小相等。在高分辨率无极化图像的各个极化单元中,通过估算出来的无极化强度数据求平均值,该平均值作为具有相应极化角度的各个极化单元的无极化强度数值,对整个阵列范围内进行同样的处理,即得到N幅低分辨率无极化图像。如图10所示,在所示的实施例中,高分辨率无极化图像的分辨率为4*4,低分辨率无极化图像的数量为4幅且分辨率为2*2。
在步骤S3处,将经过步骤S1得到的N幅低分辨率图像在步骤S2处理得到的低分辨率无极化图像的指导下,通过插值得到N幅不同极化角度的中间图像,然后再在得到的N幅中间图像中分别减去低分辨率无极化图像,即得到N幅低分辨率极化差图像;如图10所示,在所示的实施例中,得到4幅分辨率为2*2的低分辨率极化差图像
在步骤S4处,采用双线性差值、上采样的处理方法对步骤S3得到的N幅低分辨率极化差图像进行处理,得到N幅相对应的高分辨率极化差图像。如图10所示,在所示的实施例中,得到4幅分辨率为4*4的高分辨率极化差图像。
在步骤S5处,将步骤S4得到的N幅高分辨率极化差图像与步骤S2得到的高分辨率无极化图像进行求和,最终得到N幅高分辨率极化图像。如图10所示,在所示的实施例中,得到4幅高分辨率极化图像。
在根据本公开的一个实施例中,为了进一步提高高分辨率极化图像的分辨率,可以对具有极化信息的高分辨率极化图像进行超分辨率图像处理算法提高分辨率。可通过超分辨率极化成像重构算法实现超分辨成像,分辨率比现有的成像图像模式(不能获得极化信息)的提高至少4倍,分辨率可以达到毫米级别。这对识别毫米级别结构 的可疑物是非常有效的。
本领域的技术人员可以理解,上面所描述的实施例都是示例性的,并且本领域的技术人员可以对其进行改进,各种实施例中所描述的结构在不发生结构或者原理方面的冲突的情况下可以进行自由组合。
在详细说明本公开的较佳实施例之后,熟悉本领域的技术人员可清楚的了解,在不脱离随附权利要求的保护范围与精神下可进行各种变化与改变,且本公开亦不受限于说明书中所举示例性实施例的实施方式。

Claims (15)

  1. 一种用于对被检对象进行安全检查的毫米波太赫兹成像设备,其包括聚焦透镜,检测器和图形处理装置,其中
    所述聚焦透镜设置在被检对象和所述检测器之间,且被构造为将被检对象自发辐射或反射回来的毫米波太赫兹波聚焦在所述检测器上;
    检测器,所述检测器包括天线阵列和探测器阵列,其中天线阵列设置在所述探测器阵列的朝向所述聚焦透镜的一侧且设置为所述探测器阵列的天线端口,所述探测器阵列设置在所述聚焦透镜的焦平面上,且被构造为将所述天线阵列接收的毫米波太赫兹波转化为被检对象的极化图像;以及
    所述图形处理装置设置于所述探测器阵列的远离所述天线阵列的一侧,且被构造为处理所述极化图像以对被检对象进行识别分类。
  2. 根据权利要求1所述的毫米波太赫兹成像设备,其中,所述天线阵列包括多个接收天线,所述多个接收天线中的每个接收天线具有特定的极化方向。
  3. 根据权利要求2所述的毫米波太赫兹成像设备,其中,所述探测器阵列包括多个感波单元,多个感波单元的数量与多个接收天线的数量相同,所述探测器阵列上的每个感波单元的位置与所述天线阵列上的每个接收天线的位置相对应。
  4. 根据权利要求1所述的毫米波太赫兹成像设备,其中,所述天线阵列为一维天线阵列,所述探测器阵列为一维探测器阵列,所述一维天线阵列包括线性排列的多个宏像素单元,其中每个宏像素单元为N*1的天线阵列,其中N为正整数,且N≥3,且每个宏像素单元包括至少N-1个不同的极化角度。
  5. 根据权利要求1所述的毫米波太赫兹成像设备,其中,所述天线阵列为二维天线阵列,所述探测器阵列为二维探测器阵列,所述二维天线阵列包括在二维平面上排列的多个宏像素单元,其中每个宏像素单元为M 1*M 2的天线阵列,其中M 1,M 2为正整数,且M 1,M 2≥2,且每个宏像素单元包括至少N-1个不同的极化角度,其中N=M 1*M 2
  6. 根据权利要求4或5所述的毫米波太赫兹成像设备,其中,每个宏像素单元的N个接收天线包括如下方式中的至少一种:N个线极化接收天线;N-1个线极化接收天线和一个圆极化接收天线。
  7. 根据权利要求6所述的毫米波太赫兹成像设备,其中,所述N个线极化接收天线的极化角度分别为Deg1、Deg2、Deg3、...DegN,其中
    Figure PCTCN2019110408-appb-100001
    其中i为小于等于N的正整数。
  8. 根据权利要求6所述的毫米波太赫兹成像设备,其中,所述N-1个线极化接收天线的极化角度分别为Deg1、Deg2、Deg3、...DegN-1,其中
    Figure PCTCN2019110408-appb-100002
    Figure PCTCN2019110408-appb-100003
    其中i为小于等于N-1的正整数;
    其中,圆极化包括左旋圆极化和右旋圆极化中的至少一种。
  9. 根据权利要求1所述的毫米波太赫兹成像设备,还包括毫米波太赫兹辐射源,其用于向被检对象辐射毫米波太赫兹波。
  10. 根据权利要求1所述的毫米波太赫兹成像设备,其中,所述天线阵列为一维阵列,所述探测器阵列为一维阵列,所述毫米波太赫兹成像设备还包括设置在被检对象和聚焦透镜之间的光路中的可旋转扫描反射镜。
  11. 根据权利要求10所述的毫米波太赫兹成像设备,其中,所述可旋转扫描反射镜能够旋转,从而在第一旋转角度下将被检对象上的第一部位成像在一维探测器阵列的第一特定感波单元上,且在与第一旋转角度不同的第二旋转角度下将被检对象上的与第一部位不同的第二部位成像在一维探测器阵列的与第一感波单元不同的第二感波单元上。
  12. 一种使用根据权利要求1所述的毫米波太赫兹成像设备进行物体识别分类的方法,包括:
    通过所述聚焦透镜,使得被检对象自发辐射或反射回来的毫米波太赫兹波被所述天线阵列接收且聚焦在所述探测器阵列上;
    通过所述探测器阵列,将所述天线阵列接收的毫米波太赫兹波转化为被检对象的极化图像;
    利用所述图像处理装置处理所述极化图像以获得高分辨率极化图像;
    基于获得的高分辨率极化图像,利用自动识别算法进行物体识别分类。
  13. 根据权利要求12所述的物体识别分类的方法,其中,所述接收天线阵列和所述探测器阵列均是二维阵列。
  14. 根据权利要求13所述的物体识别分类的方法,其中,
    所述天线阵列包括多个宏像素单元,每个宏像素单元包括N个接收天线,N个接收天线具有至少N-1个极化角度,所述探测器阵列包括与接收天线数量相等且位置对应的感波单元,其中N为大于等于4的正整数,
    其中,利用所述图像处理装置处理所述极化图像以获得高分辨率极化图像的步骤包括:
    S1:在探测器阵列获得的极化图像中,从多个感波单元所对应的像素点中提取N幅低分辨率极化图像,每幅低分辨率极化图像具有一个极化角度且包括具有一个相同极化角度的所有像素点;
    S2:估算出极化阵列中极化角度位置处像素的无极化强度数据,得到一幅高分辨率无极化图像,高分辨率无极化图像的分辨率与天线阵列的大小相等,以及
    在高分辨率无极化图像的各个极化单元中,通过估算出来的无极化强度数据求平均值,该平均值作为具有相应极化角度的各个极化单元的无极化强度数值,对整个阵列范围内进行同样的处理,即得到N幅低分辨率无极化图像;
    S3:将经过步骤S1得到的N幅低分辨率图像在步骤S2处理得到的低分辨率无极化图像的指导下,通过插值得到N幅不同极化角度的中间图像,然后再在得到的N幅中间图像中分别减去低分辨率无极化图像,即得到N幅低分辨率极化差图像;
    S4:采用双线性差值、上采样的处理方法对步骤S3得到的N幅低分辨率极化差 图像进行处理,得到N幅相对应的高分辨率极化差图像;以及
    S5:将步骤S4得到的N幅高分辨率极化差图像与步骤S2得到的高分辨率无极化图像进行求和,最终得到N幅高分辨率极化图像。
  15. 根据权利要求14所述的物体识别分类的方法,其中,
    利用所述图像处理装置处理所述极化图像以获得高分辨率极化图像的步骤还包括:
    S6:针对在步骤S5中得到的具有极化信息的高分辨率极化图像进行超分辨率图像处理算法提高分辨率。
PCT/CN2019/110408 2018-12-29 2019-10-10 毫米波太赫兹成像设备及物体识别分类方法 WO2020134338A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811654183.3 2018-12-29
CN201811654183.3A CN109471195B (zh) 2018-12-29 2018-12-29 毫米波太赫兹成像设备及物体识别分类方法

Publications (1)

Publication Number Publication Date
WO2020134338A1 true WO2020134338A1 (zh) 2020-07-02

Family

ID=65678278

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/110408 WO2020134338A1 (zh) 2018-12-29 2019-10-10 毫米波太赫兹成像设备及物体识别分类方法

Country Status (2)

Country Link
CN (1) CN109471195B (zh)
WO (1) WO2020134338A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114296040A (zh) * 2021-12-27 2022-04-08 中国电子科技集团公司第十四研究所 一种极化切换雷达抗主瓣干扰优化方法

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109471195B (zh) * 2018-12-29 2023-11-10 同方威视技术股份有限公司 毫米波太赫兹成像设备及物体识别分类方法
CN113008907B (zh) * 2019-12-19 2024-04-12 华为技术有限公司 太赫兹感知系统和太赫兹感知阵列
CN113126174B (zh) * 2019-12-30 2022-10-04 清华大学 被动式安检设备及其光学装置
WO2021237506A1 (zh) * 2020-05-27 2021-12-02 深圳力策科技有限公司 一种无源纳米天线阵列接收器及三维成像系统
CN113777605A (zh) * 2021-09-10 2021-12-10 哈尔滨工业大学 一种被动毫米波三维成像系统及其成像方法
CN114324345B (zh) * 2021-11-01 2024-01-12 清华大学深圳国际研究生院 一种材料成像方法、装置、终端设备及存储介质

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2003102518A2 (en) * 2002-05-31 2003-12-11 New Jersey Institute Of Technology Terahertz imaging system and method
CN104932029A (zh) * 2015-06-12 2015-09-23 上海理工大学 一种主动式太赫兹人体安检系统装置及调整方法
CN105068127A (zh) * 2015-07-20 2015-11-18 华中科技大学 一种基于极化毫米波辐射的获取目标表面方位信息的方法
CN106019275A (zh) * 2016-07-06 2016-10-12 深圳市太赫兹科技创新研究院有限公司 毫米波成像系统及安检系统
CN106093937A (zh) * 2016-05-18 2016-11-09 中国电子科技集团公司第四十研究所 一种微波毫米波人体安检系统及安检方法
CN207426169U (zh) * 2017-10-20 2018-05-29 中国工程物理研究院电子工程研究所 一种用于太赫兹和毫米波人体安检仪的双极化天线阵列
CN109471194A (zh) * 2018-12-29 2019-03-15 同方威视技术股份有限公司 毫米波太赫兹成像设备及物体识别分类方法
CN109471195A (zh) * 2018-12-29 2019-03-15 同方威视技术股份有限公司 毫米波太赫兹成像设备及物体识别分类方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5227800A (en) * 1988-04-19 1993-07-13 Millitech Corporation Contraband detection system
FR2972094A1 (fr) * 2011-07-28 2012-08-31 Commissariat Energie Atomique Capteur d'image terahertz multidirectionnel
CN207764388U (zh) * 2018-01-23 2018-08-24 烟台太赫兹光电科技有限公司 主动式连续波太赫兹人体隐藏物检测装置
CN209784559U (zh) * 2018-12-29 2019-12-13 同方威视技术股份有限公司 毫米波太赫兹成像设备

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2003102518A2 (en) * 2002-05-31 2003-12-11 New Jersey Institute Of Technology Terahertz imaging system and method
CN104932029A (zh) * 2015-06-12 2015-09-23 上海理工大学 一种主动式太赫兹人体安检系统装置及调整方法
CN105068127A (zh) * 2015-07-20 2015-11-18 华中科技大学 一种基于极化毫米波辐射的获取目标表面方位信息的方法
CN106093937A (zh) * 2016-05-18 2016-11-09 中国电子科技集团公司第四十研究所 一种微波毫米波人体安检系统及安检方法
CN106019275A (zh) * 2016-07-06 2016-10-12 深圳市太赫兹科技创新研究院有限公司 毫米波成像系统及安检系统
CN207426169U (zh) * 2017-10-20 2018-05-29 中国工程物理研究院电子工程研究所 一种用于太赫兹和毫米波人体安检仪的双极化天线阵列
CN109471194A (zh) * 2018-12-29 2019-03-15 同方威视技术股份有限公司 毫米波太赫兹成像设备及物体识别分类方法
CN109471195A (zh) * 2018-12-29 2019-03-15 同方威视技术股份有限公司 毫米波太赫兹成像设备及物体识别分类方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114296040A (zh) * 2021-12-27 2022-04-08 中国电子科技集团公司第十四研究所 一种极化切换雷达抗主瓣干扰优化方法
CN114296040B (zh) * 2021-12-27 2024-04-23 中国电子科技集团公司第十四研究所 一种极化切换雷达抗主瓣干扰优化方法

Also Published As

Publication number Publication date
CN109471195A (zh) 2019-03-15
CN109471195B (zh) 2023-11-10

Similar Documents

Publication Publication Date Title
WO2020134338A1 (zh) 毫米波太赫兹成像设备及物体识别分类方法
CN209784559U (zh) 毫米波太赫兹成像设备
US10879281B2 (en) Image capture device, method of capturing image with the same, and irradiation device
US8791851B2 (en) Hybrid millimeter wave imaging system
CN105227815B (zh) 一种被动式单像素望远成像方法
US20200012119A1 (en) Reducing glare for objects viewed through transparent surfaces
CN109471194B (zh) 毫米波太赫兹成像设备及物体识别分类方法
CN108540691B (zh) 相机装置和用于拍摄至少两种波段光线的方法
Sinclair et al. Outdoor passive millimetre wave security screening
CN210534350U (zh) 毫米波太赫兹成像设备
CN109633775B (zh) 毫米波/太赫兹波成像设备对人体或物品进行检测的方法
KR100829215B1 (ko) 은닉물체 감지를 위한 합성 영상 생성 시스템
WO2020047993A1 (zh) 一种安检设备及其成像方法
Al-Shoukry An automatic hybrid approach to detect concealed weapons using deep learning
RU2702617C2 (ru) Устройство получения изображений и соответствующий способ получения изображений
CN109471196B (zh) 毫米波太赫兹成像设备及物体识别分类方法
Trichopoulos et al. Imaging performance of a THz focal plane array
JP2007124510A (ja) 電磁波イメージング装置
CN113126173A (zh) 被动式安检设备及其接收天线单元
CN113126174B (zh) 被动式安检设备及其光学装置
JPH0927927A (ja) ミリ波・赤外撮像装置
US20170160210A1 (en) Imaging apparatus for monitoring objects
Wilson et al. Polarization difference imaging for millimeter-wave in a desert environment
Harmer et al. Passive millimetre-wave imaging with a planar diffraction antenna
You et al. Quasi-Optical Dielectric Lens Design for D-Band Passive Millimeter-Wave imager

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: 19906522

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: 19906522

Country of ref document: EP

Kind code of ref document: A1