CN106525244A - Infrared fusion visual detection system - Google Patents

Abstract

The invention discloses a fusion vision detection system based on optical fiber bundle tomography and infrared multi-eye vision. The system includes an optical fiber bundle tomography subsystem, an infrared vision subsystem, a gigabit switch and a computer. The Gigabit switch forms the Gigabit local area network with the computer, the optical fiber bundle tomography subsystem and the infrared vision subsystem. The network interface is connected. The computer controls the optical fiber bundle tomography subsystem and the infrared vision subsystem through the gigabit local area network, and obtains the processing results of the optical fiber bundle tomography subsystem and the infrared image data output by the infrared vision subsystem. The beneficial effect of the present invention is that the MV is used to reconstruct the three-dimensional infrared surface topography of the target, distinguish optically thin and optically thick regions, and obtain OFBT boundary constraints; OFBT reconstructs the internal distribution of optically thin regions; the information fusion of the two solves the three-dimensional of complex objects The challenge of thermal imaging inspection.

Description

An infrared fusion vision detection system

technical field

The invention relates to an infrared fusion vision detection system, in particular to a fusion vision detection system based on optical fiber bundle tomography and infrared multi-eye vision.

Background technique

Infrared thermal imaging technology is a technology that uses infrared detectors to convert invisible infrared radiation into visible images. In the early stages of its development, infrared unit sensors are often used in combination with optical-mechanical scanning devices to obtain target infrared images. Since the 1980s, the international community began to study and apply infrared focal plane array (IRFPA) as a sensor in a thermal imager, including cooled and uncooled IRFPA, and quantum well (QWIP), quantum dot (QDIP), etc. New detectors based on quantum effects, etc. The third-generation IRFPA has made progress in the imaging area, pixel scale and multi-color imaging in the world. For example, Teledyne of the United States provided the HgCdTe IRFPA with a pixel size of 4096×4096 for NASA’s Webb Space Telescope in 2011. ; Another 1024 × 1024 pixel QWIP medium and long wave IRFPA and four-color IRFPA have also been successfully developed.

Although high-resolution IRFPA improves the resolution of thermal images, conventional infrared thermal imaging methods can only obtain the thermal image of the target local imaging plane, which provides limited information. At present, there is an urgent need for thermal imaging technology that can provide accurate target information in aerospace, national defense and military, and various industrial sectors. Since the actual target mostly includes part of the optically thick surface or is partly surrounded by external obstacles composed of optically thick materials, the infrared stereo detection of the target has become a huge problem, and the measurement method and means of the infrared volume thermal image of the target are urgently needed.

To solve this problem, the present invention proposes an infrared fusion vision method, which fuses information from Optical Fiber Bundle Tomography in Free Space (OFBT for short) and Multi-view Vision (MV for short). In order to realize the three-dimensional thermal imaging of the target to be measured and calculate the three-dimensional distribution of physical parameters such as emissivity, transmittance, and temperature.

Contents of the invention

The purpose of the present invention is to provide a fusion vision detection system based on optical fiber bundle tomography and infrared multi-eye vision. Using MV to reconstruct the blackbody passband radiation intensity distribution in the optically thick area of the target surface, and using the coordinates of the boundary points of different regions of the target surface reconstructed by MV and the corresponding passband blackbody radiation intensity as OFBT constraints, high-precision inversion of the target optically thin The passband blackbody radiation intensity field distribution corresponding to each point in the area, so as to realize the three-dimensional thermal imaging of the target to be measured and calculate its three-dimensional thermal physical quantity distribution.

The technical scheme of the present invention is realized in this way, the hardware system based on the fusion visual detection of optical fiber bundle tomography and infrared multi-eye vision mainly consists of a test platform, an optical fiber bundle tomography subsystem, an infrared vision subsystem, a gigabit switch, a computer composition;

Wherein, the optical fiber bundle tomography subsystem includes n OFBs (that is, Optical fiber bundles, optical fiber bundles), and n is an integer greater than or equal to 4, that is, from the first OFB, the second OFB, the third OFB to the nth OFB One OFB; n tripods, n spatial filters; n passband filters; n long-wave device arrays and their corresponding analog amplifiers and digital signal processing circuits;

Each OFB is composed of p optical fiber sensing units uniformly distributed, and p is the square of an integer greater than or equal to 10. Each optical fiber sensing unit is composed of a sensing head, an input coupler, a sensing fiber, an output coupler, and a long-wave unit device. A total of p long-wave unit devices in p optical fiber sensing units belonging to one OFB are arranged in order to form a long-wave device array. The outer diameter of the sensor head of the optical fiber sensing unit is equivalent to the inner diameter of the positioning hole of the spatial filter, and the total sensor heads in the root optical fiber sensing unit belonging to one OFB are taken as a group and loaded into the space filter in sequence The center of the sensor head is provided with a collimation hole, which allows the light to enter the sensor head along a straight line; a passband filter of the same size is installed in front of the spatial filter, and the light entering all the optical fiber sensing units of the OFB is controlled. Pass-band filtering; each OFB is fixed on the pan/tilt of the tripod, and its position can be adjusted conveniently; the sensing fiber is connected to the sensor head through the input coupler, and connected to the long-wave unit device through the output coupler; the long-wave device array The sensing signal can be amplified by the analog amplifier, and then processed by the digital signal processing circuit, and the processing result is sent out through the RJ45 network interface of the processing end;

The infrared vision subsystem includes m infrared vision modules, m is an integer greater than or equal to 4; there are m translation guide rails and a rotating guide rail on the test platform; m infrared vision modules are installed on the m translation guide rails , can perform one-dimensional translation along the translation guide rail and rotate around the rotation guide rail, and can be fixed after reaching the expected space and angle position; the infrared vision module mainly includes an infrared zoom lens and an infrared area array, and the infrared area array control and drive circuit has RJ45 network interface on the sensor side, through which infrared image data can be transmitted;

The Gigabit switch forms the Gigabit local area network with the computer, the optical fiber bundle tomography subsystem and the infrared vision subsystem. The network interface is connected. The computer controls the optical fiber bundle tomography subsystem and the infrared vision subsystem through the Gigabit LAN, and obtains the processing results of the optical fiber bundle tomography subsystem and the infrared image data output by the infrared vision subsystem, thereby realizing computer, optical fiber bundle Gigabit high-speed data transmission between the tomography subsystem and the infrared vision subsystem.

The steps of the fusion visual detection method based on optical fiber bundle tomography and infrared multi-eye vision are as follows:

(1) Calibration of OFB passband radiation using a black body furnace

An optical fiber sensing unit in each OFB detects an infrared radiation, and the sensing optical fiber is an infrared optical fiber (the hollow-core thermal infrared element can transmit 8-14 μm thermal infrared radiation with low loss) and is coupled with a long-wave unit device by an output coupler to realize communication. With optical fiber bundle tomographic data acquisition, the received radiation is converted into analog voltage, which becomes digital after amplification and A/D conversion.

The total of n multiplied by p optical fiber sensing units of each long-wave unit device position of n OFBs corresponds to the passband blackbody radiation intensity I _{t, the relationship between j} and the digital quantity must be calibrated in advance, I _t is the passband blackbody Radiation intensity, j is the serial number of the optical fiber sensing unit to which the long-wave unit device to be calibrated belongs, and the method adopted is:

Select the optical fiber bundle tomographic detection wavelength range of 8-14 μm, adjust the temperature of the blackbody furnace to a certain temperature T, and calculate the passband blackbody radiation intensity I _t corresponding to the temperature T according to the Planck blackbody radiation formula (that is, the radiation curve at the blackbody temperature T Area in the range of 8-14 μm). At a distance L away from the radiation cavity of the blackbody furnace, a fiber optic sensing unit to be calibrated is fixedly placed so that the end face of the sensing head is aligned and perpendicular to the center of the radiation cavity, and the output digital value of the fiber sensing unit is measured as D. Record the above set of parameter values (L,I _t ,D). Adjust the temperature T of the blackbody furnace, record another set of parameter values (L, I _t , D), after completing the calibration of the entire blackbody furnace temperature range, change L, and repeat the above steps to achieve different distances and different passband blackbody radiation intensities Next, the optical fiber sensing unit outputs digital calibration. Since there are individual differences in the long-wave unit devices of each optical fiber sensing unit, it is necessary to calibrate each optical fiber sensing unit to reduce system errors. Finally, the relationship between the passband black body radiation intensity I t,j corresponding to the passband blackbody radiation intensity I _t,j and the output digital quantity D of n multiplied by p optical fiber sensing units at different detection distances L is finally completed I _t,j =f ₁ (L,D ,j), to establish the OFB passband radiation database.

(2) Use a black body furnace for infrared vision passband radiation calibration

In order to realize the fusion of optical fiber bundle tomography and infrared multi-eye vision information, that is, to provide the boundary constraint conditions for optical fiber bundle tomographic reconstruction, it is necessary to solve the conversion problem between the gray level of the passband infrared image (8-14 μm) and the radiation intensity of the passband blackbody, namely Perform infrared vision passband radiation calibration.

The relationship between the infrared image composed of pixel gray value G and the passband blackbody radiation intensity I _t,i of each infrared area array output by the m infrared vision modules must be calibrated in advance, and I _t is the passband blackbody radiation intensity , i is the serial number of the infrared vision module to be calibrated, and the calibration method is similar to step (1):

Adjust the temperature of the blackbody furnace to a certain temperature T, and calculate the passband blackbody radiation intensity I _t corresponding to the temperature T according to the Planck blackbody radiation formula (that is, the area of the radiation curve at the blackbody temperature T in the range of 8-14 μm). At a distance L away from the radiation cavity of the black body furnace, an infrared vision module to be calibrated is fixedly placed, and the infrared zoom lens of the infrared vision module is adjusted to accurately focus on the radiation cavity of the black body furnace. Read the average gray value G of the radiation cavity part of the blackbody furnace in the infrared image output by the infrared array of the infrared vision module, and record the above-mentioned set of parameter values (L, I _t , G). Adjust the temperature T of the blackbody furnace, record another set of parameter values (L, I _t , G), after completing the calibration of the entire blackbody furnace temperature range, change L, and repeat the above steps to achieve different distances and different passband blackbody radiation intensities Next, the infrared image output by the infrared vision module constitutes the calibration of the pixel gray value G. Since there are individual differences in the infrared arrays of each infrared vision module, calibration of all infrared arrays can reduce system errors. Finally, a total of m infrared arrays in m infrared vision modules are completed at different detection distances L, and the output infrared image (8-14 μm) is composed of pixel gray value G and passband blackbody radiation intensity I _t,i The relationship I _t,i =f ₂ (L,G,i), establishes a multi-eye infrared vision passband radiation database.

(3) MV and OFBT data collection

Place the target to be tested on the test platform, and place n OFBs fixedly by the mounting plate on the tripod head at different latitude and longitude angles in the spherical space (equal spherical radius R). At the same time, m infrared vision modules are installed on m translation guide rails 8, and perform one-dimensional translation along the translation guide rails and rotate around the rotation guide rails until the angles between the translation guide rails are equal, and the m infrared vision modules After the distance between the module and the center of the spherical space is equal to R, it can be fixed, and the focus of the infrared zoom lens of all infrared vision modules is adjusted to R;

The sensor head of each OFB has a parallel collimation hole structure, which not only ensures the linear signal acquisition but also meets the real-time requirements. A total of p sensing heads in the p optical fiber sensing unit belonging to an OFB are taken as a group, and are installed in the positioning holes of the spatial filter in sequence and fixed. The outer diameter of the sensing head is equal to the diameter of the positioning hole, and the center of the positioning hole The horizontal and vertical distances between them are equal, which can meet the spatial resolution requirements for testing the target to be tested.

Carry out the fusion vision test of the target to be tested. The gigabit switch forms a Gigabit local area network with the computer, the optical fiber bundle tomography subsystem and the infrared vision subsystem. The computer obtains the output of the infrared vision subsystem through the Gigabit LAN. The infrared image data of m paths, all the pixel grayscales G of each infrared image are converted against the multi-eye infrared vision passband radiation database I _t,i =f ₂ (R,G,i) to obtain m frames with Infrared radiation image characterized by blackbody passband radiation intensity I _t .

At the same time, the digital signal processing circuit controls the working power of each long-wave unit device of all n times p optical fiber sensing units by controlling the electronic switch, so as to realize the instantaneous start-up of all long-wave unit devices in the form of electronic shutter to collect the target to be measured The spatial multi-point passband radiation intensity signal entering each sensor head, and the output terminal of each long-wave unit device is connected to the peak hold circuit, so that the collected spatial multipoint passband radiation intensity signal is converted into a voltage signal and locked in the peak hold circuit . After being amplified by an analog amplifier, and then processed by a digital signal processing circuit, all data is converted from an analog signal to a digital signal to obtain a digital quantity D, and then compared with the OFB passband radiation database I _t,j =f ₁ (R,D ,j), to obtain the original data I _t of the passband radiation intensity of multiple points in the target space to be measured.

(4) Fusion data processing

For the m infrared radiation images in the computer characterized by the black body passband radiation intensity I _t , the Harris operator is used to detect the corners in different regions, based on the different spatial and frequency domain characteristics of the infrared passband optically thin and optically thick regions (such as grayscale, histogram, texture, amplitude-frequency distribution, etc.) for region segmentation, and an image matching algorithm based on epipolar constraints to perform infrared radiation image feature corner matching on m frames, reconstructing the surface topography of the target to be measured, and The three-dimensional coordinates of the boundary of different areas on the surface and the intensity value of the blackbody radiation in the passband. According to the reconstructed boundaries of different areas on the surface of the target to be measured, the target to be measured is spatially segmented, that is, divided into optically thin and optically thick regions of the infrared passband. For the optically thick region, the reconstructed surface topography of the target to be measured is directly used , that is, to complete the work of infrared multi-eye vision; for optically thin areas, enter the following three-dimensional reconstruction of optical fiber bundle tomography space:

Taking the three-dimensional coordinates of the boundaries of different areas of the surface reconstructed by infrared multi-eye vision and the passband blackbody radiation intensity values as constraints, the original passband radiation intensity of multiple points in the target space to be measured obtained by the digital signal processing circuit in step (2) The data is divided, the radiation intensity data in the optically thick area is discarded, and the original data of the passband radiation intensity in the optically thin area is used to perform the inversion calculation of the fiber bundle emission optical tomography. Through tomographic calculation, the passband radiation intensity of each point in the internal space of the optically thin area of the target to be measured can be obtained, so as to be compared with the surface topography and passband of the target to be measured 1 reconstructed by infrared multi-eye vision in the optically thick area The radiation intensity is fused together to complete the reconstruction of the spatial three-dimensional passband radiation intensity I _t distribution of the entire target to be measured (including optically thin and optically thick areas), and based on this, the target to be measured (including optically thin and optically thick The distribution of physical quantities such as three-dimensional temperature, pressure, and particle number density in the area) space, and the three-dimensional distribution results are displayed on the computer in real time to complete the entire fusion visual detection.

The beneficial effect of the present invention is to overcome the limitations of single OFBT and MV target 3D thermal image detection, use MV to reconstruct the 3D infrared surface topography of the target, distinguish optically thin and optically thick regions, and obtain OFBT boundary constraints; OFBT reconstructs optically thin regions Internal distribution; the information fusion of the two solves the problem of 3D thermal image detection of complex objects.

Description of drawings

Fig. 1 is a schematic diagram of the present invention, in the figure: 1—target to be measured; 2—spherical space; 3—tripod; 4—spatial filter; 5—positioning hole; 6—first OFB; 7—pass band filter; 8—translational guide rail; 9—second OFB; 10—third OFB; 11—rotary guide rail; 12—infrared array; 13— RJ45 network interface at the sensing end; 14—infrared zoom lens; 15—the nth OFB; 16—test platform; 17—collimation hole; 18—sensing head; 19—sensing optical fiber; 20 ——Input coupler; 21—Long-wave device array; 22—Analog amplifier; 23—Digital signal processing circuit; 24—RJ45 network interface at the processing end; 25—Twisted pair; 26—Gigabit switch; 27—computer; 28—infrared vision module; 29—output coupler; 30—long wave unit device.

Note: OFB is Optical fiber bundle; RJ45 is Registered Jack 45 data transmission interface; n is the total number of OFB.

detailed description

The hardware system structure of the fusion visual detection based on optical fiber bundle tomography and infrared multi-eye vision is shown in Figure 1. The hardware system mainly consists of a test platform 16, an optical fiber bundle tomography subsystem, an infrared vision subsystem, a gigabit switch 26, and a computer. 27 composition;

Wherein the optical fiber bundle tomography subsystem includes n OFBs, namely from the first OFB 6, the second OFB9, the third OFB10 to the nth OFB 15 (in this embodiment, n is 4); n Tripod 3, n spatial filters 4; n passband filters 7; n long-wave device arrays 21 and their corresponding analog amplifiers 22 and digital signal processing circuits 23;

Each OFB is composed of p optical fiber sensing units evenly distributed. Each optical fiber sensing unit is composed of a sensing head 18 , an input coupler 20 , a sensing fiber 19 , an output coupler 29 , and a long-wave unit device 30 . A total of p long-wave unit devices 30 in p optical fiber sensing units belonging to one OFB are arranged in sequence to form a long-wave device array 21 . The outer diameter of the sensing head 18 of the optical fiber sensing unit is equivalent to the inner diameter of the positioning hole 5 of the spatial filter 4, and a total of p sensing heads 18 in the p optical fiber sensing unit belonging to an OFB are taken as a group, in order Put it into the positioning hole 5 of the spatial filter 4 and fix it. There is a collimation hole 17 in the center of the sensing head 18, allowing light to enter the sensing head 18 along a straight line; the spatial filter 4 is equipped with a pass-band filter of the same size 7 (its passband is 8～14 μ m in the present embodiment), carry out passband filtering to the light that enters all optical fiber sensing units of OFB; Each OFB is all fixed on the cloud platform of tripod 3, and its position can be adjusted conveniently; The sensing optical fiber 19 is connected to the sensor head 18 through the input coupler 20, and connected to the long-wave unit device 30 through the output coupler 29; the sensing signal of the long-wave device array 21 can be amplified by the analog amplifier 22, and then passed through the digital signal processing circuit 23 is processed, and the processing result is transmitted outwards through the RJ45 network interface 24 of the processing end;

The infrared vision subsystem includes m infrared vision modules 28; m translation guide rails 8 and a rotating guide rail 11 are arranged on the test platform 16; m infrared vision modules 28 are installed on the m translation guide rails 8, which can be Perform one-dimensional translation along the translation guide rail 8 and rotate around the rotation guide rail 11, and can be fixed after reaching the expected space and angle position; the infrared vision module 28 mainly includes an infrared zoom lens 14 and an infrared area array 12, and the infrared area array 12 controls and There is an RJ45 network interface 13 at the sensing end on the drive circuit, through which infrared image data can be transmitted outwards;

Gigabit switch 26 forms Gigabit local area network with computer 27, optical fiber bundle tomography subsystem and infrared vision subsystem, and all processing end RJ45 network interface 24 and sensing end RJ45 network interface 13 and computer 27 network interfaces are all through twisted pair 25 is connected with the network interface of gigabit switch 26. The computer 27 controls the optical fiber bundle tomography subsystem and the infrared vision subsystem through the gigabit local area network, and obtains the processing results of the optical fiber bundle tomography subsystem and the infrared image data output by the infrared vision subsystem, thereby realizing computer 27, Gigabit high-speed data transmission between the fiber optic bundle tomography subsystem and the infrared vision subsystem.

The steps of the fusion visual detection method based on optical fiber bundle tomography and infrared multi-eye vision are as follows:

(1) Calibration of OFB passband radiation using a black body furnace

An optical fiber sensing unit in each OFB detects an infrared radiation line, and the sensing optical fiber 19 adopts an infrared optical fiber (the hollow-core thermal infrared element can transmit 8-14 μm thermal infrared radiation with low loss) and couples the long-wave unit device through the output coupler 29 30 (this embodiment adopts the domestic OTP538U unit device, and its spectral response range is 8-14 μm) to realize the tomographic data acquisition of the pass-band fiber optic bundle, and convert the received radiation into an analog voltage, which becomes a digital quantity after amplification and A/D conversion .

The total of n OFBs multiplied by the pass-band blackbody radiation intensity I _{t corresponding to the position of each long-wave unit device 30 of the p root optical fiber sensing unit, the relationship between j} and the digital quantity must be calibrated in advance, and I _t is the pass-band Blackbody radiation intensity, j is the serial number of the optical fiber sensing unit to which the long-wave unit device 30 belongs to to be calibrated (in this embodiment, n is 4, p is 900, and the range of j is 1 to 3600), and the method adopted is:

Select the optical fiber bundle tomographic detection wavelength range of 8-14 μm, adjust the temperature of the blackbody furnace to a certain temperature T, and calculate the passband blackbody radiation intensity I _t corresponding to the temperature T according to the Planck blackbody radiation formula (that is, the radiation curve at the blackbody temperature T Area in the range of 8-14 μm). At a distance L away from the radiation cavity of the blackbody furnace, a fiber sensing unit to be calibrated is fixedly placed so that the end face of the sensor head 18 is aligned and perpendicular to the center of the radiation cavity, and the output digital value of the fiber sensing unit is measured to be D. Record the above set of parameter values (L,I _t ,D). Adjust the temperature T of the blackbody furnace, record another set of parameter values (L, I _t , D), after completing the calibration of the entire blackbody furnace temperature range, change L, and repeat the above steps to achieve different distances and different passband blackbody radiation intensities Next, the optical fiber sensing unit outputs digital calibration. Since there are individual differences in the long-wave unit device 30 of each optical fiber sensing unit, it is necessary to calibrate each optical fiber sensing unit to reduce system errors. Finally, the relationship between the passband black body radiation intensity I t,j corresponding to the passband blackbody radiation intensity I _t,j and the output digital quantity D of n multiplied by p optical fiber sensing units at different detection distances L is finally completed I _t,j =f ₁ (L,D ,j), to establish the OFB passband radiation database.

(2) Use a black body furnace for infrared vision passband radiation calibration

In order to realize the fusion of optical fiber bundle tomography and infrared multi-eye vision information, that is, to provide the boundary constraint conditions for optical fiber bundle tomographic reconstruction, it is necessary to solve the conversion problem between the gray level of the passband infrared image (8-14 μm) and the radiation intensity of the passband blackbody, namely Perform infrared vision passband radiation calibration.

Each infrared array 12 in the m infrared vision modules 28 (this embodiment uses a vanadium oxide uncooled infrared focal plane array, and its working wavelength range is 8-14 μm) outputs an infrared image composed of pixel grayscale values G and Passband blackbody radiation intensity I _{t, the relationship of i} must be pre-calibrated, I _t is the passband blackbody radiation intensity, and i is the serial number of the infrared vision module 28 to be calibrated (in this embodiment, m gets 4, i's The range is 1 to 4), the calibration method is similar to step (1):

Adjust the temperature of the blackbody furnace to a certain temperature T, and calculate the passband blackbody radiation intensity I _t corresponding to the temperature T according to the Planck blackbody radiation formula (that is, the area of the radiation curve at the blackbody temperature T in the range of 8-14 μm). At a distance L away from the radiation cavity of the black body furnace, an infrared vision module 28 to be calibrated is fixedly placed, and the infrared zoom lens 14 of the infrared vision module 28 is adjusted to accurately focus on the radiation cavity of the black body furnace. Read the average gray value G of the radiation cavity part of the blackbody furnace in the infrared image output by the infrared array 12 of the infrared vision module 28, and record the above-mentioned set of parameter values (L, I _t , G). Adjust the temperature T of the blackbody furnace, record another set of parameter values (L, I _t , G), after completing the calibration of the entire blackbody furnace temperature range, change L, and repeat the above steps to achieve different distances and different passband blackbody radiation intensities Next, the infrared image output by the infrared vision module 28 constitutes the calibration of the pixel gray value G. Since there are individual differences in the infrared area arrays 12 of each infrared vision module 28, all infrared area arrays 12 are calibrated to reduce system errors. Finally, a total of m infrared arrays 12 in m infrared vision modules 28 are completed at different detection distances L, and the output infrared images (8-14 μm) are composed of pixel gray value G and passband blackbody radiation intensity I _t, The relation of _i is I _t,i =f ₂ (L,G,i), and the multi-eye infrared vision passband radiation database is established.

(3) MV and OFBT data collection

The target 1 to be tested is placed on the test platform 16, and n OFBs are fixedly placed on the mounting plate on the tripod 3 at different latitude and longitude angles in the spherical space 2 (equal spherical radius R). At the same time, m infrared vision modules 28 are installed on m translation guide rails 8, and perform one-dimensional translation along the translation guide rails 8 and rotate around the rotation guide rails 11 until the angles between the translation guide rails 8 are equal, and M infrared vision modules 28 can be fixed after being equal to R from the spherical space 2 sphere center distances, and the focus of the infrared zoom lens 14 of all infrared vision modules 28 is adjusted to R;

The sensing head 18 of each OFB has a structure of parallel collimation holes, which not only guarantees linear signal collection but also meets real-time requirements. A total of p sensing heads 18 in the p optical fiber sensing units belonging to one OFB are used as a group, and are loaded into the positioning holes 5 of the spatial filter 4 in sequence and fixed. The outer diameter of the sensing heads 18 is the same as the diameter of the positioning holes 5 Equal (both 1 mm in this embodiment), the horizontal and vertical distances between the centers of the positioning holes 5 are equal (the distance is 2 mm in this embodiment), which can meet the spatial resolution requirements of the target 1 to be tested.

Carry out the fusion vision test of the target to be tested 1, the Gigabit switch 26 forms the Gigabit local area network with the computer 27, the optical fiber bundle tomography subsystem and the infrared vision subsystem, and the computer 27 obtains the relevant test data output by the infrared vision subsystem through the Gigabit local area network For the m-channel infrared image data under different angles of the target 1, the grayscale G of all pixels of each infrared image is converted against the multi-eye infrared vision passband radiation database I _t,i =f ₂ (R,G,i) , to obtain m infrared radiation images characterized by blackbody passband radiation intensity I _t .

At the same time, the digital signal processing circuit 23 controls the working power of each long-wave unit device 30 of all n multiplied by p optical fiber sensing units by controlling the electronic switch, so as to realize the instantaneous start-up of all long-wave unit devices 30 in an electronic shutter mode to collect The target 1 to be measured enters the spatial multipoint passband radiation intensity signal of each sensor head 18, and the output terminal of each long-wave unit device 30 is connected to a peak hold circuit at the same time, so that the collected spatial multipoint passband radiation intensity signal is converted into a voltage signal Latched in the peak hold circuit. After being amplified by the analog amplifier 22, and then processed by the digital signal processing circuit 23, all data are converted from analog signals to digital signals to obtain the digital quantity D, and then compared with the OFB passband radiation database I _{t, j} =f ₁ (R ,D,j), to obtain the original data I _t of the passband radiation intensity of multiple points in the target 1 space.

(4) Fusion data processing

For the m infrared radiation images in the computer 27 characterized by the black body passband radiation intensity I _t , the Harris operator is used to detect the corners in different regions, based on the different image space and frequency domains of the infrared passband optically thin and optically thick regions Features (such as grayscale, histogram, texture, amplitude-frequency distribution, etc.) are used for region segmentation, and the image matching algorithm based on epipolar line constraints is used to match the feature corners of infrared radiation images for m frames, and the surface morphology of the target 1 to be measured is reconstructed , as well as the three-dimensional coordinates of the boundary of different areas on the surface and the radiation intensity value of the blackbody in the passband. According to the reconstructed boundaries of different regions on the surface of the target to be measured 1, the target to be measured 1 is spatially segmented, that is, divided into optically thin and optically thick regions of the infrared passband. For the optically thick region, the reconstructed target to be measured 1 is directly used Surface topography, that is, to complete the work of infrared multi-eye vision; for optically thin areas, enter the following three-dimensional reconstruction of optical fiber bundle tomography space:

Using the three-dimensional coordinates of the boundaries of different areas of the surface reconstructed by infrared multi-eye vision and the passband blackbody radiation intensity values as constraints, the passband radiation of the multi-point passband radiation of the target to be measured 1 space obtained by processing the digital signal processing circuit 23 in step (2) The intensity raw data is divided, the radiation intensity data in the optically thick region is discarded, and the passband radiation intensity raw data in the optically thin region is used for the inversion calculation of optical tomography of the optical fiber bundle emission. Through tomographic calculation, the passband radiation intensity of each point in the internal space of the optically thin area of the target 1 to be measured can be obtained, so as to be consistent with the surface topography and passband of the target 1 to be measured reconstructed by infrared multi-eye vision in the optically thick area. The three-dimensional passband radiation intensity I t distribution of the entire object to be measured 1 (including the optically thin area and the optically thick area) is fused together to complete the reconstruction of the spatial three-dimensional passband radiation intensity I _t distribution of the target to be measured 1 (including the optically thin area). and optically thick region) spatial distribution of three-dimensional temperature, pressure, particle number density and other physical quantities, the three-dimensional distribution results are displayed on the computer 27 in real time to complete the entire fusion visual detection.

Claims (1)

1. it is a kind of chromatographed based on fibre bundle and infrared multi-vision visual fusion visual detection system, the system includes test platform (16), fibre bundle chromatography subsystem, infrared vision subsystem, gigabit switch (26) and computer (27), it is characterised in that：

Described fibre bundle chromatography subsystem includes n OFB, and n is the integer more than or equal to 4, i.e., from first OFB (6), the Two OFB (9), the 3rd OFB (10), until n-th OFB (15)；N tripod (3), n spatial filter (4)；N Band pass filter (7)；N long wave device array (21) and its corresponding analogue amplifier (22) and digital signal processing circuit (23)；

Each OFB is made up of the equally distributed Fibre Optical Sensor unit of p roots, p be the integer more than or equal to 10 square, every optical fiber Sensing unit is by sensing head (18), input coupler (20), sensor fibre (19), output coupler (29), long wave unit component (30) constitute.Belong to and amount to p long wave unit component (30) arrangement group in order in the p root Fibre Optical Sensor units of an OFB Into a long wave device array (21)；The external diameter of the sensing head (18) of Fibre Optical Sensor unit and the location hole of spatial filter (4) (5) internal diameter quite, belongs to the p sensing head (18) altogether in the p root Fibre Optical Sensor units of an OFB as one group, by suitable During sequence puts into the location hole (5) of spatial filter (4) and fixed, sensing head (18) center is provided with collimating aperture (17), it is allowed to light Sensing head (18) is entered along straight line；Band pass filter (7) of the assembling with size before spatial filter (4), to into all light of OFB The light of fine sensing unit carries out passband optical filtering；Each OFB is fixed on the head of tripod (3), and its position is conveniently adjusted； Sensor fibre (19) is connected with sensing head (18) by input coupler (20), by output coupler (29) and long wave unit device Part (30) is connected；The transducing signal of long wave device array (21) can Jing analogue amplifiers (22) be amplified, then Jing digital signals Process circuit (23) is processed, and result is outwards transmitted by processing end RJ45 network interfaces (24)；

Described infrared vision subsystem includes m infrared vision module (28), and m is the integer more than or equal to 4；Test platform (16) there are m translation guide rail (8) and a rotary rail (11) on；M infrared vision module (28) are led installed in this m translation On rail (8), one-dimensional translation can be made along translation guide rail (8) and be rotated around rotary rail (11), to after expected space and angle position Can fix；Infrared vision module (28) mainly include infrared zooming lens (14) and infrared surface battle array (12), infrared surface battle array (12) control There is sensor ends RJ45 network interface (13) on system and drive circuit, infrared picture data can be outwards transmitted by the interface；

Computer (27), fibre bundle chromatography subsystem are constituted gigabit with infrared vision subsystem by described gigabit switch (26) The network of LAN, all process end RJ45 network interfaces (24) and sensor ends RJ45 network interface (13) and computer (27) Interface is all passed through twisted-pair feeder (25) and is connected with the network interface of gigabit switch (26).Computer (27) passes through the gigabit LAN To fibre bundle chromatography subsystem be controlled with infrared vision subsystem, and obtain fibre bundle chromatography subsystem result and With infrared vision subsystem output infrared picture data, so as to realize computer (27), fibre bundle chromatography subsystem with it is infrared Gigabit level high speed data transfer between vision subsystem.