Distributed Polarization Detection System and Method Based on Position-Regulated Target Reconstruction

Abstract

The present invention discloses a distributed polarization detection system and method based on position-regulated target reconstruction, belonging to the fields of distributed polarization collaborative detection and optoelectronic imaging. The distributed polarization detection system includes an illumination device, a multi-parameter auxiliary detection stage, and a long-wave infrared polarization detection device. The long-wave infrared polarization detection device has three sub-polarization detection devices. This method is implemented based on the distributed polarization detection system. Through the distributed polarization detection system based on position-regulated target reconstruction, polarization information in three different directions can be collected at the same time, improving the recognition ability of camouflaged / similar targets at a single angle and avoiding errors caused by time delay. The polarization information obtained by reasonably distributing the detection positions complements each other, and the polarization images are fused with each other, effectively extracting the target signal from the scattered light and analyzing the polarization characteristic differences between the interfering light and the target reflected light, which helps to improve the resolution, contrast, and detection distance of polarization imaging.

Description

Technical Field

[0001] The present invention belongs to the field of distributed polarization cooperative detection and optoelectronic imaging, and in particular, relates to a distributed polarization detection system and method based on position-controlled target reconstruction. Background Art

[0002] Polarization detection imaging technology is to highlight the target from the environment, thereby improving the resolution of different targets and environments, but it is still affected by some harsh environments and the limitations of single detection. With the growing demand for detection, how to effectively detect the environment, accurately identify the physical characteristics of the target, and achieve collaborative detection in the process of polarization detection is a technical problem that needs to be solved urgently.

[0003] In actual polarization imaging detection, the dynamic changes of the detection environment and the need for real-time detection require that the detection device be changed according to the actual environment, and a single detection is difficult to achieve an ultra-large dynamic range and high spatial resolution. Summary of the invention

[0004] In view of this, the purpose of the present invention is to solve the problem of distributed polarization imaging collaborative detection of multiple targets in the long-wave infrared band, and to propose a distributed polarization detection system and method based on position-controlled target reconstruction.

[0005] The technical solution adopted by the present invention to achieve the above-mentioned purpose is:

[0006] According to the first aspect of the present invention, a distributed polarization detection system based on position-regulated target reconstruction is proposed. The distributed polarization detection system includes an illumination device, a multi-parameter auxiliary detection stage, and a long-wave infrared polarization detection device, and the illumination device, the multi-parameter auxiliary detection stage, and the long-wave infrared polarization detection device are respectively connected to an image display and control system; the illumination device is equipped with an artificial light source simulating the sun and a light source adjustment mechanism for adjusting the position and / or angle of the artificial light source; the multi-parameter auxiliary detection stage includes a stage main body, a first photodetector, a graphene heating plate, a heat insulation board, and a second photodetector. The first photodetector is fixedly attached to the central position of the upper surface of the stage main body; the graphene heating plate is arranged above the first photodetector at a set spacing; the heat insulation board is arranged around the first photodetector and the graphene heating plate; the second photodetector is an annular photodetector. The second photodetector is placed above the heat insulation board. The central area of the second photodetector is used as the placement area for the target, and the bottom surface of the target is attached to the upper surface of the graphene heating plate; the long-wave infrared polarization detection device has three sub-polarization detection devices, and the three sub-polarization detection devices are dispersedly distributed on one side of the multi-parameter auxiliary detection stage. Each of the three sub-polarization detection devices has a polarization detection component and a polarization detection adjustment mechanism for adjusting the position and / or angle of the polarization detection component. The polarization detection component includes a long-wave infrared polarization detection module and a laser gas detection photoelectric module. The laser gas detection photoelectric module includes a laser, a collimating and beam-expanding optical unit, and a converging lens. The lasers carried by the three sub-polarization detection devices are a 671nm laser, a 1064nm laser, and a carbon dioxide laser respectively; the long-wave infrared polarization detection module includes a long-wave infrared quarter-wave plate, a long-wave infrared polarizer, and a long-wave infrared camera arranged in sequence along the light propagation direction.

[0007] Further, the light source adjustment mechanism includes a light source adjustment mechanism base, a first telescopic mechanical lower arm, a first rotatable upper arm, and a first wrist rotating shaft. The first telescopic mechanical lower arm is installed on the light source adjustment mechanism base; both ends of the first rotatable upper arm are rotatably connected to the first telescopic mechanical lower arm and the first wrist rotating shaft through bearings respectively; an artificial light source is installed on the first wrist rotating shaft. The light source adjustment mechanism base, the first telescopic mechanical lower arm, the first rotatable upper arm, and the first wrist rotating shaft are all equipped with separate drive motors.

[0008] Further, the spacing between the first photodetector and the graphene heating plate is 5 cm.

[0009] Further, the long-wave infrared quarter-wave plate is installed on a first electrically controlled high-speed rotating wheel; the long-wave infrared polarizer is installed on a second electrically controlled high-speed rotating wheel.

[0010] Furthermore, the polarization detection adjustment mechanism includes a polarization detection adjustment mechanism base, a rotating shaft, a second telescopic mechanical lower arm, a telescopic rotating middle arm, a second rotatable upper arm, and a second wrist rotating shaft, which are connected in sequence from bottom to top. A polarization detection component is installed on the second wrist rotating shaft. Separate drive motors are installed on the polarization detection adjustment mechanism base, the rotating shaft, the second telescopic mechanical lower arm, the telescopic rotating middle arm, the second rotatable upper arm, and the second wrist rotating shaft.

[0011] According to the second aspect of the present invention, a distributed polarization detection method based on position-regulated target reconstruction is proposed. This method is implemented using the aforementioned distributed polarization detection system and specifically includes:

[0012] Step 1: Place the target above and in contact with the graphene heating plate on the multi-parameter auxiliary detection stage.

[0013] Step 2: Turn on the 671nm laser, 1064nm laser, and carbon dioxide laser. The laser beams emitted by the 671nm laser, 1064nm laser, and carbon dioxide laser are emitted as parallel beams after passing through their respective corresponding collimating and beam-expanding optical units. When there is no haze gas in the air, the parallel beams will be converged by the converging lens onto the center of the second photodetector; when there is haze gas in the air, the laser will scatter when it encounters haze particles and will finally be captured by each detection unit in the second photodetector.

[0014] Step 3: Turn off the 1064nm laser and the carbon dioxide laser. Use the 671nm laser as the emission end. After passing through the collimating and beam-expanding optical unit and the converging lens, it irradiates the target surface. The first photodetector is used as the receiving end. If the target is opaque, the first photodetector below the graphene heating plate does not receive a signal; if the target is transparent, after the light is birefringed by the transparent target, it passes through the graphene heating plate and the first photodetector receives the optical signal.

[0015] Step 4: Turn off the 671nm laser, turn on the artificial light source of the lighting device, and adjust the light source adjustment mechanism to determine the incident zenith angle of the artificial light source and ensure that the light is incident on the target surface.

[0016] Step 5: Turn on the polarization detection component of the long-wave infrared polarization detection device and adjust the polarization detection adjustment mechanism to determine the detection zenith angle and relative azimuth angle of the polarization detection component; when blocking the artificial light source, keep the detection zenith angle and relative azimuth angle unchanged and adjust the polarization detection adjustment mechanism to avoid the blockage.

[0017] Step 6: Adjust the positions of the long-wave infrared quarter-wave plate and the long-wave infrared polarizer to generate linearly polarized light at 0°, 45°, 90°, 135°, left-handed circularly polarized light, and right-handed circularly polarized light. Then, let the light of these six polarization states enter the long-wave infrared camera respectively. Change the relative azimuth angle and repeat the experiment to complete the polarization detection for the full relative azimuth angle ranging from 0° to 360°;

[0018] Step 7: Change the detection zenith angle and repeat the experiment in Step 6 to complete the detection for the full detection zenith angle ranging from 0° to 90° and the full relative azimuth angle ranging from 0° to 360° at the starting incident zenith angle;

[0019] Step 8: Change the incident zenith angle and repeat Steps 6 and 7 to complete the polarization detection imaging experiment for the polarization characteristics of the unheated target in the full wavelength band;

[0020] Step 9: Power on the graphene heating plate to heat the target. When the target temperature reaches the set temperature, stop powering on the graphene heating plate, and the heat insulation board maintains the target temperature;

[0021] Step 10: Repeat Steps 6 to 9. Use the polarization detection component again to perform polarization detection imaging on the polarization state of the target at the set temperature, and read the image information of the polarization detection imaging;

[0022] Step 11: After Steps 1 to 10, the initial acquisition of the target polarization image information by the three sub-polarization detection devices is completed. The target images detected by the three sub-polarization detection devices respectively cover a part of the entire scene and there are overlapping areas between the three images, forming three groups of polarization state image datasets. Each group of polarization state image datasets includes polarization state images of linearly polarized light at 0°, 45°, 90°, 135°, left-handed circularly polarized light, and right-handed circularly polarized light. Then, fuse the six polarization state images in each group into a polarization degree image, and establish three groups of polarization degree image datasets X1, X2, and X3;

[0023] Step 12: Use the position coordinates, overlapping areas, and polarization degree change information provided by the three groups of polarization degree image datasets X1, X2, and X3 as the input layer. Based on the particle swarm optimization-based BP neural network optimization algorithm, find the size of the overlapping area and the optimal position, determine the optimal coordinate positions of the three sub-polarization detection devices, and obtain the best distribution positions of the three sub-polarization detection devices;

[0024] Step 13: Initialize the lighting device and the long-wave infrared polarization detection device, that is, set the incident zenith angle of the artificial light source, the detection zenith angle and the relative azimuth angle of the polarization detection component to 0°, and distribute the optimal coordinate positions of the three sub-polarization detection devices according to the optimal distribution positions obtained by optimizing in Step 12; set the target initial temperature to room temperature; vary the relative azimuth angle at intervals of 10°, with a range of 0° to 360°; vary the detection zenith angle at intervals of 10°, with a range of 0° to 90°; vary the incident zenith angle at intervals of 10°, with a range of 0° to 90° for detection; the three sub-polarization detection devices acquire images; when the temperature increases by 15 °C on the basis of the initial temperature, repeat the detection of the target with the relative azimuth angle varying at intervals of 10°, with a range of 0° to 360°; the detection zenith angle varying at intervals of 10°, with a range of 0° to 90°; the incident zenith angle varying at intervals of 10°, with a range of 0° to 90°; until the temperature reaches 105 °C to end the detection. Thus, the acquisition of the distributed target polarization images is completed;

[0025] Step 14: After the acquisition of the distributed target polarization images is completed, use the multi-view sparse scene reconstruction method to calculate the polarization state information and intensity information of the target scene for the images of the three sub-polarization detection devices at different temperatures; on this basis, based on the visual saliency mapping, perform multi-dimensional information enhancement and feature fusion on the original source image data, and finally reconstruct the polarization fusion image;

[0026] Among them, the original source image refers to the initial imaging image received by the detection.

[0027] Furthermore, the process of fusing the six polarization state images in each group into a polarization degree image is as follows: Based on the wavelet transform in the multi-scale image fusion algorithm, the low-frequency coefficients are fused using the principal component analysis transformation method, and the high-frequency coefficients are fused using the regional feature energy fusion rule. Finally, the polarization degree image fused from the six polarization state images in each group is obtained through the inverse wavelet transform.

[0028] Through the above design scheme, the present invention can bring the following beneficial effects: The distributed polarization detection system and method based on position-regulated target reconstruction can collect polarization information in three different directions at the same time, improve the recognition ability of camouflaged / similar targets at a single angle, and avoid errors caused by time delay. The polarization information obtained by reasonably distributing the detection positions complements each other, and the polarization images are fused with each other, effectively extracting the target signal from the scattered light to analyze the polarization characteristic differences between the interfering light and the target reflected light, which helps to improve the resolution, contrast and detection distance of polarization imaging. Description of the Drawings

[0029] The accompanying drawings described herein are used to provide a further understanding of the present invention, and constitute a part of this application. The schematic embodiments of the present invention and their descriptions are used to understand the present invention, and do not unduly limit the present invention. In the drawings:

[0030] Figure 1 It is a schematic structural diagram of a distributed polarization detection system for target reconstruction based on position regulation;

[0031] Figure 2 It is a schematic top view distribution diagram of three sub-polarization detection devices;

[0032] Figure 3 It is a schematic structural diagram of an illumination device;

[0033] Figure 4 It is a schematic structural diagram of a multi-parameter auxiliary detection stage;

[0034] Figure 5 It is a schematic structural diagram of a long-wave infrared polarization detection device;

[0035] Figure 6 It is a schematic structural diagram of a polarization detection component;

[0036] Figure 7 It is a schematic workflow diagram of a distributed polarization detection method for target reconstruction based on position regulation.

[0037] The marks in the figure are as follows: 1 is the illumination device; 11 is the artificial light source; 12 is the base of the light source adjustment mechanism; 13 is the first retractable mechanical lower arm; 14 is the first rotatable upper arm; 15 is the first wrist rotating shaft; 2 is the multi-parameter auxiliary detection stage; 21 is the main body of the stage; 22 is the first photodetector; 23 is the graphene heating plate; 24 is the heat insulation board; 25 is the second photodetector; 26 is the target; 27 is the IV and AD conversion circuit; 28 is the monitor; 29 is the temperature controller, 30 is the DC regulated power supply; 3 is the long-wave infrared polarization detection device; 3-1 is the sub-polarization detection device A; 3-2 is the sub-polarization detection device B; 3-3 is the sub-polarization detection device C; 31 is the polarization detection component; 311 is the long-wave infrared polarization detection module; 312 is the collimating and beam expanding optical unit; 313 is the converging lens; 314 is the 671nm laser; 315 is the 1064nm laser; 316 is the carbon dioxide laser; 317 is the long-wave infrared quarter-wave plate; 318 is the long-wave infrared polarizer; 319 is the long-wave infrared camera; 320 is the first electronically controlled high-speed rotating wheel; 321 is the second electronically controlled high-speed rotating wheel; 32 is the base of the polarization detection adjustment mechanism; 33 is the rotating shaft; 34 is the second retractable mechanical lower arm; 35 is the retractable and rotatable middle arm; 36 is the second rotatable upper arm; 37 is the second wrist rotating shaft. Detailed implementation manners

[0038] In order to make the objectives, features, and advantages of the present invention more obvious and understandable, the following describes the technical solutions of the present invention clearly and completely in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the present invention is not limited by the following embodiments, and the specific implementation manners can be determined according to the technical solutions of the present invention and the actual situation. In order to avoid confusing the essence of the present invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0039] As Figure 1 and Figure 2 shown, the distributed polarization detection system based on position-regulated target reconstruction proposed by the present invention includes an illumination device 1, a multi-parameter auxiliary detection stage 2, and a long-wave infrared polarization detection device 3. Finally, the image display and control system at the software end receives and stores images, and controls the illumination device 1, the multi-parameter auxiliary detection stage 2, and the long-wave infrared polarization detection device 3. The image display and control system belongs to the prior art and will not be described in detail here. The illumination device 1 is equipped with an artificial light source 11 that simulates the sun and a light source adjustment mechanism for adjusting the position and / or angle of the artificial light source 11. The light source adjustment mechanism includes a light source adjustment mechanism base 12 that rotates horizontally through a rotary motor (not shown in the drawings), a first telescopic mechanical lower arm 13 that is driven to vertically expand and contract by a linear motor (not shown in the drawings), a first rotatable upper arm 14 and a first wrist rotating shaft 15 that are respectively controlled by their corresponding servo motors (not shown in the drawings). The first telescopic mechanical lower arm 13 is installed on the light source adjustment mechanism base 12; both ends of the first rotatable upper arm 14 are rotatably connected to the first telescopic mechanical lower arm 13 and the first wrist rotating shaft 15 through bearings; the artificial light source 11 is fixed on the first wrist rotating shaft 15. The first telescopic mechanical lower arm 13 in the illumination device 1 is used to adjust the height of the light rays of the artificial light source 11, the first rotatable upper arm 14 is used to roughly adjust the zenith angle of incidence of the light rays of the artificial light source 11 and make the light rays irradiate on the surface of the target 26, and the first wrist rotating shaft 15 is used to precisely adjust the zenith angle of incidence of the light rays of the artificial light source 11. The entire illumination device 1 can achieve simulated illumination with different heights and different zenith angles of incidence through coordinated cooperation.

[0040] The multi-parameter assisted detection stage 2 includes a stage main body 21, a first photodetector 22, a graphene heating plate 23, a heat insulation board 24, and a second photodetector 25. The first photodetector 22 is fixedly attached to the central position of the upper surface of the stage main body 21. The graphene heating plate 23 is fixed 5 cm above the first photodetector 22, and the distance between the graphene heating plate 23 and the first photodetector 22 is 5 cm, so that an air layer forms a thermal resistance to ensure that the operating temperature of the first photodetector 22 is stably within a safe range, avoiding aging or signal drift of the first photodetector 22 caused by thermal stress. The heat insulation board 24 is arranged around the first photodetector 22 and the graphene heating plate 23 to prevent other components from being damaged due to the increase in temperature of the graphene heating plate 23. The second photodetector 25 is an annular photodetector. The second photodetector 25 is placed above the heat insulation board 24. The central area of the second photodetector 25 is used as the placement area for the target 26, and the bottom surface of the target 26 is attached to the upper surface of the graphene heating plate 23. The graphene heating plate 23 realizes electro-thermal conversion heating by driving the migration and energy dissipation of carriers through the electric field after being energized, and is externally connected to a temperature controller 29, which is powered by a DC regulated power supply 30. The specific structures of the temperature controller 29 and the DC regulated power supply 30, as well as the connection method with the graphene heating plate 23, all belong to the prior art and will not be elaborated in detail here. The second photodetector 25 measures the scattered light intensities at different angles, obtains the scattered light intensity distribution information and converts it into an electrical signal, and converts the electrical signal into a digital signal through an IV and AD conversion circuit 27, which is displayed on a monitor 28 to complete the haze detection work. When the multi-parameter assisted detection stage 2 is used for polarization detection of the target 26, it heats the target 26, assists in detecting the experimental gas environment and whether the target 26 is transparent. If the target 26 is transparent, the light is double-refracted by the transparent target 26. Due to the high light transmittance of the graphene heating plate 23, the light passes through the graphene heating plate 23, and the first photodetector 22 receives the optical signal.

[0041] The long-wave infrared polarization detection device 3 has three sub-polarization detection devices, and the three sub-polarization detection devices are dispersedly distributed on one side of the multi-parameter auxiliary detection stage 2. Each of the three sub-polarization detection devices has a polarization detection component 31 and a polarization detection adjustment mechanism for adjusting the position and / or angle of the polarization detection component 31. The polarization detection component 31 includes a long-wave infrared polarization detection module 311 and a laser gas detection optoelectronic module. The laser gas detection optoelectronic module includes a laser, a collimating and beam-expanding optical unit 312, and a converging lens 313. The collimating and beam-expanding optical unit 312 is an optical system that combines collimation and beam-expansion functions, and is used to convert a diverging light beam into a parallel light beam and expand its diameter. The specific structure of the collimating and beam-expanding optical unit 312 belongs to the prior art and will not be described in detail separately here. The lasers carried by the three sub-polarization detection devices are a 671 nm laser 314, a 1064 nm laser 315, and a carbon dioxide laser 316 respectively. For the convenience of description, the three sub-polarization detection devices are respectively named sub-polarization detection device A 3-1, sub-polarization detection device B 3-2, and sub-polarization detection device C 3-3. Among them, the 671 nm laser 314 is carried in the sub-polarization detection device A 3-1, the 1064 nm laser 315 is carried in the sub-polarization detection device B 3-2, and the carbon dioxide laser 316 is carried in the sub-polarization detection device C 3-3. The polarization detection adjustment mechanism includes a polarization detection adjustment mechanism base 32, a rotating shaft 33, a second telescopic mechanical lower arm 34, a telescopic and rotating middle arm 35, a second rotatable upper arm 36, and a second wrist rotating shaft 37, which are connected in sequence from bottom to top. The polarization detection adjustment mechanism base 32 and the rotating shaft 33 adjust their horizontal orientations through their respective corresponding rotating motors. The second telescopic mechanical lower arm 34 is driven by a linear motor to perform vertical telescoping. The telescopic and rotating middle arm 35 is controlled by the coordinated operation of a rotating motor and a linear motor. The second rotatable upper arm 36 and the second wrist rotating shaft 37 are respectively controlled by their respective corresponding servo motors. The polarization detection component 31 is installed on the second wrist rotating shaft 37. The polarization detection adjustment mechanism base 32 is used to determine the position of the entire sub-polarization detection device, and the rotating shaft 33 drives the entire sub-polarization detection device to rotate through rotation; the second telescopic mechanical lower arm 34 controls the height through telescoping. The telescopic and rotating middle arm 35, the second rotatable upper arm 36, and the second wrist rotating shaft 37 can rotate 180° around the axis, and cooperate with each other to position the polarization detection component 31.The long-wave infrared polarization detection module 311 includes a long-wave infrared quarter-wave plate 317, a long-wave infrared polarizer 318, and a long-wave infrared camera 319 arranged in sequence along the light propagation direction. The long-wave infrared quarter-wave plate 317 is installed on the first electronically controlled high-speed rotating wheel 320; the long-wave infrared polarizer 318 is installed on the second electronically controlled high-speed rotating wheel 321. The electronically controlled high-speed rotating wheel is a mechanical device that realizes high-speed and precise rotation through an electronic control system, and is widely used in fields such as industrial automation, scientific research instruments, and optical systems. It belongs to the prior art and will not be elaborated here in detail. It should be understood that the terms "first" and "second" are only used for descriptive purposes, and the features defined with "first" and "second" do not represent any order, quantity, or importance, but are only used to distinguish different components. The sub-polarization detection device A 3-1, the sub-polarization detection device B 3-2, and the sub-polarization detection device C 3-3 respectively use a 671nm laser 314, a 1064nm laser 315, and a carbon dioxide laser 316 as laser light sources, and then convert the light emitted by the light sources into parallel light through their respective corresponding collimating and beam-expanding optical units 312, and irradiate the second photodetector 25 of the multi-parameter auxiliary detection stage 2 through the converging lens 313. When the laser meets the particles in the gas, scattering will occur, and the gas environment can be measured by measuring the light intensity distribution after the laser scattering through the second photodetector 25.

[0042] The multi-parameter auxiliary detection stage 2 is located between the lighting device 1 and the long-wave infrared polarization detection device 3. The first photodetector 22 cooperates with the sub-polarization detection device A 3-1 to detect whether the target 26 is transparent; the second photodetector 25 cooperates with the sub-polarization detection device A 3-1, the sub-polarization detection device B 3-2, and the sub-polarization detection device C 3-3 to detect the haze concentration in the gas; the graphene heating plate 23 and the heat insulation board 24 provide a temperature change environment for the target 26.

[0043] The long-wave infrared polarization detection device 3 performs polarization detection on the hemispherical space where the target 26 is located, and solves the problem that the light emitted by the lighting device 1 is blocked by the long-wave infrared polarization detection device 3 during the detection process through the adjustment of the robotic arm; the long-wave infrared polarization detection device 3 combines the advantages of distributed polarization detection to solve the polarization detection problem of different types of targets in the long-wave infrared band.

[0044] The lighting device 1, the multi-parameter auxiliary detection stage 2, and the long-wave infrared polarization detection device 3 constitute a distributed infrared polarization detection system. The detection images in the long-wave infrared polarization detection device 3 are displayed through the image display and control system, and the drive motors in the lighting device 1 and the long-wave infrared polarization detection device 3 are controlled to drive the mechanical components to designated positions for detection, further promoting the coordinated cooperation of the entire distributed polarization detection system based on position-regulated target reconstruction, realizing full-automatic long-wave infrared band distributed polarization detection in a haze environment, and performing image processing on the subsequent polarization images, position regulation of the three sub-polarization detection devices, and reconstruction of target polarization information.

[0045] To achieve the purpose of distributed polarization detection, the long-wave infrared polarization detection device 3 has three sub-polarization detection devices. The three sub-polarization detection devices are set for collaborative detection, and they are randomly and dispersedly distributed around the detected target 26. First, the experimental gas environment is monitored by the different-band lasers carried by the three sub-polarization detection devices and preliminary polarization detection is carried out. According to the image information obtained from the preliminary polarization detection, the optimal distribution positions and the size of the overlapping area of the three sub-polarization detection devices are calculated. Then, the three sub-polarization detection devices are re-detected according to the optimal distribution positions, and the obtained images are reconstructed to obtain the polarization image after the final position-regulated target reconstruction.

[0046] The present invention proposes to obtain images through preliminary polarization detection using the designed distributed polarization detection system based on position-regulated target reconstruction, solve the image information using the BP neural network optimization algorithm based on particle swarm optimization and optimize the position layout of the three sub-polarization detection devices, perform position regulation and re-detect the target 26 to obtain image information, and reconstruct the target image information using the multi-view sparse scene reconstruction method. The calculation result of visual saliency mapping is used as the weight, and iterative guided filtering enhances the multi-dimensional information and fuses the features of the details of the image area, and finally obtains the reconstructed target image and calculates its polarization state information and intensity information.

[0047] The specific distributed polarization detection method based on position-regulated target reconstruction includes the following steps:

[0048] Step 1: Place the target 26 on the graphene heating plate 23 of the multi-parameter auxiliary detection stage 2 and fit it with the graphene heating plate 23 to facilitate uniform heating.

[0049] Step 2: Turn on the 671 nm laser 314, 1064 nm laser 315, and carbon dioxide laser 316. The 671 nm laser 314, 1064 nm laser 315, and carbon dioxide laser 316 emit laser beams, which form parallel beams after passing through their respective collimating and beam expanding optical units 312. When there is no haze gas in the air, the parallel beams will be converged by the converging lens 313 onto the central ring of the second photodetector 25. When there is haze gas in the air, the laser will scatter when it encounters haze particles, and finally the intensity distribution of the scattered light at different angles will be captured by the second photodetector 25, converting the optical signal into an electrical signal. The IV and AD conversion circuit 27 converts the electrical signal into a digital signal, which is finally displayed on the monitor 28 to achieve accurate measurement of the haze concentration.

[0050] Step 3: Turn off the 1064 nm laser 315 and carbon dioxide laser 316. Using the 671 nm laser 314 as the transmitting end, after passing through the collimating and beam expanding optical unit 312 and the converging lens 313, it irradiates the surface of the target 26. The first photodetector 22 is used as the receiving end. By detecting the transmittance and reflectivity of the target 26, it is determined whether the target 26 is transparent. If the target 26 is opaque, the light is reflected by the surface of the target 26, and there is no optical signal at the first photodetector 22 below the graphene heating plate 23. If the target 26 is transparent, after the light is birefringed by the transparent target 26, due to the high light transmittance of the graphene heating plate 23, the light passes through the graphene heating plate 23, and the first photodetector 22 receives the optical signal.

[0051] Step 4: Turn off the 671 nm laser 314, turn on the artificial light source 11 of the lighting device 1, adjust the first retractable mechanical lower arm 13 so that the light evenly irradiates the target 26, and adjust the first rotatable upper arm 14 and the first wrist rotating shaft 15 to determine the incident zenith angle of the artificial light source 11 and ensure that the light is incident on the surface of the target 26.

[0052] Step 5: Turn on the polarization detection component 31 of the long-wave infrared polarization detection device 3. The second retractable mechanical lower arm 34 and the retractable and rotatable middle arm 35 adjust the detection height. The second rotatable upper arm 36 and the second wrist rotating shaft 37 determine the detection zenith angle. The rotating shaft 33, the retractable and rotatable middle arm 35, the second rotatable upper arm 36, and the second wrist rotating shaft 37 determine the relative azimuth angle. When blocking the laser light source, keep the detection zenith angle and the relative azimuth angle unchanged, and adjust the angle between the arms to avoid blocking.

[0053] Step 6: Set the first electronically controlled high-speed rotating wheel 320 and the second electronically controlled high-speed rotating wheel 321 to control the positions of the long-wave infrared quarter-wave plate 317 and the long-wave infrared polarizer 318, so as to form linearly polarized light at 0°, 45°, 90°, 135°, left-handed circularly polarized light, and right-handed circularly polarized light. The six polarization state lights enter the long-wave infrared camera 319, change the relative azimuth angle in intervals of 10°, repeat the experiment, and complete the polarization detection of the full relative azimuth angle ranging from 0° to 360°.

[0054] Step 7: Change the detection zenith angle, repeat the experiment in Step 6, and complete the detection of the full detection zenith angle ranging from 0° to 90° and the full relative azimuth angle ranging from 0° to 360° at the starting incident zenith angle.

[0055] Step 8: Change the incident zenith angle, repeat Step 6 and Step 7, and complete the full-band polarization detection imaging experiment on the polarization characteristics of the unheated target 26.

[0056] Step 9: Use the DC regulated power supply 30 to supply power to the graphene heating plate 23 to heat the target 26. When the temperature of the target 26 reaches the set temperature, the temperature controller 29 disconnects the power supply, and the heat insulation board 24 maintains the target temperature.

[0057] Step 10: Repeat Steps 6 to 9, and use the polarization detection component 31 again to perform polarization detection imaging on the polarization state of the target 26 at the set temperature, and read the image information of the polarization detection imaging.

[0058] Step 11: As Figure 7 shown, the preliminary data collection of the above Steps 1 to 10 is completed. The images of the target 26 obtained by the three sub-polarization detection devices respectively cover a part of the entire scene and there is an overlapping area between the three images, forming three groups of six polarization state (the polarization states of four linearly polarized lights at 0°, 45°, 90°, 135°, and left-handed circularly polarized light and right-handed circularly polarized light) image data sets. Based on the wavelet transform in the multi-scale image fusion algorithm, the low-frequency coefficients are fused by using the principal component analysis (Principal Component Analysis, abbreviated as PCA) transformation method, and the high-frequency coefficients are fused by using the regional feature energy fusion rule. Finally, the polarization degree images of the six polarization state images in each group are obtained through the inverse wavelet transform, and three groups of polarization degree image data sets X1, X2, and X3 are established. The multi-scale image fusion algorithm and the regional feature energy fusion rule both belong to the prior art and will not be elaborated in detail here.

[0059] Step 12: The image datasets X1, X2, and X3 provide position coordinates, overlapping regions, and polarization degree change information as the input layer. The process of finding the optimal distribution positions and overlapping regions of the three sub-polarization detection devices through images is not a linear relationship. Therefore, a non-linear change is used as the hidden layer, and the BP neural network optimization algorithm based on particle swarm optimization is used to find the size of the overlapping region and the optimal position, that is, the optimal coordinate distribution. The specific principle is as follows:

[0060] (1);

[0061] In Equation (1), is the position of the sub-polarization detection device A 3-1 at time ; is the position of the sub-polarization detection device A 3-1 at time ; is the velocity calculated according to the velocity update formula, that is:

[0062] (2);

[0063] In Equation (2), is the velocity of the sub-polarization detection device A 3-1 at time ; is the velocity of the sub-polarization detection device A 3-1 at time ; is the inertia weight, which controls the influence of the velocity of the sub-polarization detection device A 3-1; is the individual learning factor, also known as the cognitive coefficient, which affects the degree to which the sub-polarization detection device A 3-1 moves towards the optimal position; is a random number within the range of [0, 1], which increases randomness; is the historical optimal position of the sub-polarization detection device A 3-1; is the social learning factor, also known as the social coefficient, which affects the degree to which the sub-polarization detection device A 3-1 moves towards the optimal position of the distributed polarization detection system; is another random number within the range of [0, 1], which also increases randomness; is the global optimal position, which is the optimal solution found by the sub-polarization detection device A 3-1 in the position coordinates.

[0064] To ensure that the sub-polarization detection device does not exceed the hemispherical space, a constraint condition is added after the position update. Assume that the center of the hemisphere is at the origin, and a hemispherical region with an outer radius of is formed by the sub-polarization detection device A 3-1, the sub-polarization detection device B 3-2, and the sub-polarization detection device C 3-3. The upper surface of the target to be measured 26 serves as the bottom surface of the hemispherical region, that is, the x-y plane. Calculate the distance from the new position of the sub-polarization detection device A 3-1 to the origin :

[0065] (3);

[0066] If is greater than , the position of the sub-polarization detection device A 3-1 needs to be re-projected back onto the hemispherical surface. This can be achieved by normalizing the position vector of the sub-polarization detection device A 3-1 to the unit sphere and then multiplying by the radius :

[0067] (4);

[0068] If the detection position of the sub-polarization detection device A 3-1 is at the bottom of the hemispherical region (i.e., the z coordinate is negative), its z coordinate needs to be set to 0 and re-normalized to remain on the hemispherical surface. To further ensure that the detection position of the sub-polarization detection device A 3-1 does not exceed the hemispherical space, similar restrictions are imposed on the velocity vector. If the moving velocity direction of the sub-polarization detection device A 3-1 points outside the hemisphere, its velocity vector is projected back into the hemisphere.

[0069] Similarly, based on the particle swarm optimization-based BP neural network optimization algorithm, the optimal coordinate positions of the sub-polarization detection device B 3-2 and the sub-polarization detection device C 3-3 are obtained, and the fitness function and root mean square error ( ) are calculated to measure the optimization performance and used as influencing factors to perform back-optimization on the weights and thresholds of the BP neural network until the prediction error of the network reaches the minimum. The root mean square error is:

[0070] (5);

[0071] In the formula, is the number of samples; is the actual output value of the th sample; is the predicted output value of the th sample. The fitness value is defined as:

[0072] (6);

[0073] Minimize , so as to maximize the fitness value, and comprehensively consider the training error and the test error , such as:

[0074] (7);

[0075] Achieve a balance between training and generalization capabilities to obtain the optimal distribution positions of the sub-polarization detection devices, further obtain the coordinate relationships of the three sub-polarization detection devices, and calculate the size of the overlapping area in combination with the included angles between the fields of view. Use a set of image datasets to verify the optimized BP neural network and evaluate its accuracy in predicting the optimal positions of the sub-polarization detection devices.

[0076] Step 13: As Figure 7 shown, initialize the lighting device 1 and the long-wave infrared polarization detection device 3, that is, set the incident zenith angle, detection zenith angle, and relative azimuth angle of the light source to 0°. Distribute the sub-polarization detection devices according to the optimal distribution positions obtained by the above optimization; set the initial temperature of the target 26 to room temperature, that is, 25°C; vary the relative azimuth angle at intervals of 10°, with a range from 0° to 360°; vary the detection zenith angle at intervals of 10°, with a range from 0° to 90°; vary the incident zenith angle at intervals of 10°, with a range from 0° to 90° for detection, and the three sub-polarization detection devices acquire images; when the temperature increases by 15°C based on the initial temperature, repeat the detection of the relative azimuth angle varying at intervals of 10°, with a range from 0° to 360°; the detection zenith angle varying at intervals of 10°, with a range from 0° to 90°; the incident zenith angle varying at intervals of 10°, with a range from 0° to 90° for the target 26; stop the detection until the temperature reaches 105°C.

[0077] Step 14: Use the multi-view sparse scene reconstruction method to solve for the polarization state information and intensity information of the target scene from the images of the three sub-polarization detection devices obtained at different temperatures. On this basis, based on the information weighting strategy driven by the visual saliency mapping model, perform multi-dimensional information enhancement and feature fusion on the original source image data, and finally reconstruct a polarization fusion image with high information entropy, where the original source image refers to the initial imaging image received by the detection. The multi-view sparse scene reconstruction method is a technology in the field of computer vision used to recover the three-dimensional structure of a scene from images taken from multiple perspectives, which belongs to the prior art and will not be elaborated here.

[0078] First, perform scale-invariant feature transform on the images obtained by the three sub-polarization detection devices. The scale-invariant feature transform (SIFT) algorithm essentially searches for feature points in images in different scale spaces and calculates the directions of the feature points to solve problems such as the rotation, scaling, translation, image affine / projection transformation, illumination influence, and target occlusion of the target; the extracted feature points are matched between different images, and the accuracy of the matching is improved by removing mis-matched points.

[0079] The matched feature points are continuously added with new views and image points through image registration and triangulation. By calculating the homography matrix or fundamental matrix, etc., the images from different perspectives are aligned and registered. Through the triangulation method, using the projection relationship of the matched points in different perspectives, the three-dimensional coordinates of the feature points in the scene are calculated, and a sparse three-dimensional point cloud is initially constructed.

[0080] The bundle adjustment is used to optimize the poses of the long-wave infrared cameras 319 and the positions of the sparse points in the distributed polarization detection system. The bundle adjustment minimizes the reprojection error of the sparse image points in the images from different perspectives. The camera parameters and positions are iteratively adjusted to improve the reconstruction accuracy. During the optimization process, the outlier points are removed and the weights of the image points are calculated to further eliminate the noise points.

[0081] The basic image layers are weighted and superimposed to generate a preliminary reconstructed image. Through the iterative guided filter algorithm, the difference between every two iterations is calculated to extract the detail layer. The least weighted square method is used to superimpose the detail layers of each layer to further enhance the detail information of the image, and finally a high-precision reconstructed image is generated.

[0082] The iterative guided filter focuses on local detail optimization during the image reconstruction process. Therefore, the parameter adjustment or region priority of the iterative guided filter is guided by the saliency mapping, that is, the result of the visual saliency mapping is used as the weight; then the iterative guided filter is used to enhance the details of these regions.

[0083] The method based on visual saliency mapping uses the low-frequency information in the image. This method defines the saliency by the difference between one pixel and other pixels. Let be the image and be the intensity value of the pixel in the image. The saliency value of the pixel is defined as , and the definition of

[0084] (8);

[0085] where is the pixel intensity and is the number of pixels with the same intensity. Then is normalized to [0,1]. Let , and represent the visual saliency mappings of the three input images respectively. The fused basic layer is obtained through the following weighted average formula:

[0086] (9);

[0087] In formula (9), is the fused basic layer, , and are the basic layers obtained by decomposing the three infrared images detected by the sub-polarization detection device A 3-1, the sub-polarization detection device B 3-2, and the sub-polarization detection device C 3-3 respectively, , and are the corresponding weights, and these weights can be calculated through visual saliency mapping.

[0088] The detail layer fusion is performed by anisotropic diffusion processing on the three infrared images, and then the detail layer is extracted. The calculation of the weights is extended as:

[0089] (10);

[0090] wherein, is the saliency map of the th image detail layer, is the saliency map of the th image detail layer, is the corresponding fusion weight. The fusion of the final detail layer is expressed as:

[0091] (11);

[0092] wherein, , and are the detail layers of the three infrared images. The final fused image F is obtained by fusing the fused basic layer and the detail layer with assigned weights:

[0093] (12);

[0094] wherein, is the fusion weight, and finally the polarization detection fusion image of the distributed detection system based on position-regulated target reconstruction is obtained.

[0095] Obviously, the above embodiments of the present invention are merely examples for clearly explaining the present invention and are not limitations on the embodiments of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the embodiments here. Any obvious changes or modifications derived from the technical solutions of the present invention are still within the protection scope of the present invention.

Claims

1. A distributed polarization detection system based on position control target reconstruction, the distributed polarization detection system comprising an illumination device (1), a multi-parameter auxiliary detection stage (2) and a long-wave infrared polarization detection device (3), wherein the illumination device (1), the multi-parameter auxiliary detection stage (2) and the long-wave infrared polarization detection device (3) are respectively connected to an image display and control system; characterized in that: The lighting device (1) comprises an artificial light source (11) simulating the sun and a light source adjustment mechanism for adjusting the position and / or angle of the artificial light source (11); the multi-parameter auxiliary detection stage (2) comprises a stage body (21), a first photodetector (22), a graphene heating plate (23), a heat-insulating plate (24) and a second photodetector (25); the first photodetector (22) is fixedly attached to the center of the upper surface of the stage body (21); the graphene heating plate (23) is arranged above the first photodetector (22) at a set interval; the heat-insulating plate (24) is arranged on the periphery of the first photodetector (22) and the graphene heating plate (23); the second photodetector (25) is an annular photodetector; the second photodetector (25) is arranged above the heat-insulating plate (24); the central area of ​​the second photodetector (25) serves as a placement area for a target (26); and the bottom surface of the target (26) is adjacent to the top surface of the graphene heating plate (23). Surface bonding; the long-wave infrared polarization detection device (3) comprises three sub-polarization detection devices, and the three sub-polarization detection devices are dispersed on one side of the multi-parameter auxiliary detection stage (2); the three sub-polarization detection devices all comprise a polarization detection component (31) and a polarization detection adjustment mechanism for adjusting the position and / or angle of the polarization detection component (31); the polarization detection component (31) comprises a long-wave infrared polarization detection module (311) and a laser gas detection photoelectric module; the laser gas detection photoelectric module comprises a laser, a collimating beam expansion optical unit (312) and a converging lens (313); and the lasers carried in the three sub-polarization detection devices are respectively a 671 nm laser (314), a 1064 nm laser (315) and a carbon dioxide laser (316); the long-wave infrared polarization detection module (311) comprises a long-wave infrared quarter-wave plate (317), a long-wave infrared polarizer (318) and a long-wave infrared camera (319) arranged in sequence along the light propagation direction.

2. The distributed polarization detection system based on position control target reconstruction according to claim 1, characterized in that: The light source adjustment mechanism comprises a light source adjustment mechanism base (12), a first telescopic mechanical lower arm (13), a first rotatable upper arm (14) and a first wrist rotation shaft (15); the first telescopic mechanical lower arm (13) is mounted on the light source adjustment mechanism base (12); two ends of the first rotatable upper arm (14) are rotatably connected to the first telescopic mechanical lower arm (13) and the first wrist rotation shaft (15) via bearings respectively; an artificial light source (11) is mounted on the first wrist rotation shaft (15); and the light source adjustment mechanism base (12), the first telescopic mechanical lower arm (13), the first rotatable upper arm (14) and the first wrist rotation shaft (15) are all equipped with separate drive motors.

3. The distributed polarization detection system based on position control target reconstruction according to claim 1, characterized in that: The distance between the first photodetector (22) and the graphene heating plate (23) is 5 cm.

4. The distributed polarization detection system based on position control target reconstruction according to claim 1, characterized in that: The long-wave infrared quarter-wave plate (317) is mounted on a first electrically controlled high-speed rotating wheel (320); and the long-wave infrared polarizing plate (318) is mounted on a second electrically controlled high-speed rotating wheel (321).

5. The distributed polarization detection system based on position control target reconstruction according to claim 1, characterized in that: The polarization detection adjustment mechanism comprises a polarization detection adjustment mechanism base (32), a rotation axis (33), a second telescopic mechanical lower arm (34), a telescopic rotating middle arm (35), a second rotatable upper arm (36) and a second wrist rotating shaft (37) which are connected in sequence from bottom to top; a polarization detection component (31) is mounted on the second wrist rotating shaft (37); and the polarization detection adjustment mechanism base (32), the rotation axis (33), the second telescopic mechanical lower arm (34), the telescopic rotating middle arm (35), the second rotatable upper arm (36) and the second wrist rotating shaft (37) are all equipped with separate drive motors.

6. A distributed polarization detection method based on position control target reconstruction, characterized in that: The method is implemented using the distributed polarization detection system described in any one of claims 1 to 5, and specifically includes: Step 1: The target (26) is placed above the graphene heating plate (23) in the multi-parameter auxiliary detection stage (2) and is bonded thereto; Step 2: Start the 671nm laser (314), the 1064nm laser (315) and the carbon dioxide laser (316); the laser beams emitted by the 671nm laser (314), the 1064nm laser (315) and the carbon dioxide laser (316) are collimated and expanded by the corresponding optical units (312) to emit parallel beams; when there is no haze gas in the air, the parallel beams are converged by the converging lens (313) on the center of the second photodetector (25); when there is haze gas in the air, the laser light encounters haze particles and is scattered, and is finally captured by the detection units in the second photodetector (25); Step 3: Turn off the 1064 nm laser (315) and the carbon dioxide laser (316), use the 671 nm laser (314) as the transmitting end, and irradiate the target (26) surface after passing through the collimating beam expansion optical unit (312) and the converging lens (313), and use the first photodetector (22) as the receiving end. If the target (26) is opaque, the first photodetector (22) below the graphene heating plate (23) does not receive the signal; if the target (26) is transparent, the light passes through the graphene heating plate (23) after being birefringent by the transparent target (26), and the first photodetector (22) receives the light signal; Step 4: Turn off the 671 nm laser (314), turn on the artificial light source (11) of the lighting device (1), adjust the light source adjustment mechanism, determine the incident zenith angle of the artificial light source (11), and ensure that the light is incident on the surface of the target (26); Step 5: turning on the polarization detection component (31) of the long-wave infrared polarization detection device (3), adjusting the polarization detection adjustment mechanism, and determining the detection zenith angle and relative azimuth angle of the polarization detection component (31); when the artificial light source (11) is blocked, keeping the detection zenith angle and relative azimuth angle unchanged, and adjusting the polarization detection adjustment mechanism to avoid blocking; Step 6: Adjust the positions of the long-wave infrared quarter-wave plate (317) and the long-wave infrared polarizer (318) to generate 0° linear polarized light, 45° linear polarized light, 90° linear polarized light, 135° linear polarized light, left-handed circular polarized light, and right-handed circular polarized light, and the six polarization states of light enter the long-wave infrared camera (319) respectively, change the relative azimuth angle, repeat the experiment, and complete the polarization detection in the full relative azimuth angle range of 0° to 360°; Step 7: Change the detection zenith angle and repeat the experiment in step 6 to complete the detection of the full detection zenith angle ranging from 0° to 90° and the full relative azimuth angle ranging from 0° to 360° under the initial incident zenith angle; Step 8: Change the incident zenith angle, repeat steps 6 and 7, and complete the full-band polarization detection imaging experiment of the polarization characteristics of the unheated target (26); Step 9: supplying power to the graphene heating plate (23) to heat the target (26), and when the temperature of the target (26) reaches the set temperature, the power supply to the graphene heating plate (23) is stopped, and the heat insulation plate (24) maintains the temperature of the target (26); Step 10: repeating steps 6 to 9, again using the polarization detection component (31) to perform polarization detection imaging on the polarization state of the target (26) at the set temperature, and reading image information of the polarization detection imaging; Step 11: After step 1 to step 10, the three sub-polarization detection devices have completed preliminary acquisition of polarization image information of the target (26). The target (26) images detected and acquired by the three sub-polarization detection devices respectively cover a part of the entire scene and there is an overlapping area between the three images, forming three groups of polarization state image data sets, each group of polarization state image data sets includes polarization state images of 0° linear polarization light, 45° linear polarization light, 90° linear polarization light, 135° linear polarization light, left-handed circular polarization light and right-handed circular polarization light, and the six polarization state images in each group are fused into a polarization degree image, and three groups of polarization degree image data sets X1, X2, X3 are established; Step 12: The position coordinates, overlapping areas, and polarization change information provided by the three sets of polarization image data sets X1, X2, and X3 are used as the input layer, and the BP neural network optimization algorithm based on particle swarm optimization is used to find the size of the overlapping area and the optimal position, determine the optimal coordinate positions of the three sub-polarization detection devices, and obtain the optimal distribution positions of the three sub-polarization detection devices; Step 13: Initialize the lighting device (1) and the long-wave infrared polarization detection device (3), that is, set the incident zenith angle of the artificial light source (11), the detection zenith angle and the relative azimuth angle of the polarization detection component (31) to 0°, and distribute the optimal coordinate positions of the three sub-polarization detection devices according to the optimal distribution positions obtained by optimization in step 12; set the initial temperature of the target (26) to room temperature; change the relative azimuth angle in intervals of 10°, ranging from 0° to 360°; change the detection zenith angle in intervals of 10°, ranging from 0° to 90°; The zenith angle is changed at intervals of 10°, ranging from 0° to 90° for detection; three sub-polarization detection devices acquire images; when the temperature increases by 15°C on the basis of the initial temperature, the relative azimuth angle of the target (26) is repeatedly changed at intervals of 10°, ranging from 0° to 360°; the detection zenith angle is changed at intervals of 10°, ranging from 0° to 90°; the incident zenith angle is changed at intervals of 10°, ranging from 0° to 90°; the detection is terminated until the temperature reaches 105°C, at which point the acquisition of the distributed target polarization image is completed; Step 14: After the distributed target polarization image is acquired, the images of the three sub-polarization detection devices at different temperatures are used to calculate the polarization state information and intensity information of the target scene using a multi-view sparse scene reconstruction method; on this basis, based on visual saliency mapping, multi-dimensional information enhancement and feature fusion are performed on the original source image data, and finally a polarization fusion image is reconstructed; The original source image refers to the initial imaging image received by the detection.

7. The distributed polarization detection method based on position control target reconstruction according to claim 6, characterized in that: The process of fusing the six polarization state images in each group into a polarization degree image is as follows: based on the multi-scale image fusion algorithm, the low-frequency coefficients are fused using the principal component analysis transformation method using wavelet transform, and the high-frequency coefficients are fused using the regional feature energy fusion rule. Finally, the polarization degree image fused from the six polarization state images in each group is obtained through inverse wavelet transform.

Images