Open Access
Add to BibSonomy
Abstract
Laser Doppler vibrometers (LDVs) traditionally used for ground vibration sensing in laser-acoustic detection of buried objects are limited to operation from a stationary platform due to their sensitivity to the motion of the LDV itself. In order to overcome this limitation a novel Laser Multi-Beam Differential Interferometric Sensor (LAMBDIS), has been developed. The LAMBDIS allows for measurements of vibration fields with interferometric sensitivity, while having low sensitivity to the motion of the sensor itself. The LAMBDIS described in this paper is based on a digital line-scan CMOS camera and FPGA based real-time signal processing. The principle of operation of the LAMBDIS employs the interference of light reflected from different points on the object surface illuminated with an array of laser beams. The Doppler shift induced by the sensor motion is canceled out thereby allowing for measurements from a moving vehicle. The ability of the LAMBDIS to detect buried objects in real time from a moving vehicle has been confirmed in field experiments.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Laser-acoustic detection of buried objects, such as landmines, has proven itself as a technique that provides high probability of detection and low false alarm rate [1–4]. The method consists of exciting vibrations in the ground and measuring vibration of the ground surface at multiple points with a laser Doppler vibrometer (LDV) to create a vibration image of the ground surface. Vibrations of the ground in the frequency range from about 50 Hz to 300 Hz are excited by using airborne sound created by loudspeakers or seismic waves created by mechanical shakers. The interaction of a buried object with the elastic waves in the ground causes the object to vibrate. Due to the mechanical resonances and the higher mechanical compliance of the buried object compared to the neighboring soil, the vibration amplitude of the ground surface above the object is higher than the vibration amplitude of the surrounding area. Therefore, a buried object can be detected by the area with higher vibration amplitude in the vibration image of the interrogated area. The laser-acoustic technique was successfully used for detection of antitank and antipersonnel mines in field tests [1–8].
Single beam and multiple beam LDVs have been traditionally used for ground vibration sensing in laser-acoustic detection of buried objects [4–8]. However, since a LDV measures the relative velocity between the LDV and the object, the measurement results depend not only on the object vibration, but also on the motion of the LDV itself. As a result, motion of the LDV due to ambient vibration and acoustic noise can cause LDV signals to be significantly higher than, and indistinguishable from, signals caused by the object vibration. To alleviate this problem, LDVs must operate from a mechanically stable and acoustically isolated platform. Correction of the measurements for LDV vibration caused by environmental noise can be done using accelerometers mounted on an LDV [9–11].
Operation of a LDV from a moving platform imposes another challenge for buried object detection. LDV operation from a moving vehicle with laser beams looking forward induces Doppler shift in the LDV signal proportional to the speed of the vehicle. This Doppler shift caused by the vehicle motion can be several orders of magnitude, exceeding the Doppler shift caused by the ground vibration, and often exceeding the modulation bandwidth of the LDV thereby making measurements impossible. This limitation can be overcome either by increasing the processing bandwidth, or by tracking the Doppler shift of the light reflected back from the ground. Increasing the bandwidth increases the LDV noise and reduces the sensitivity. Doppler tracking has been used to compensate for the LDV motion [10]. However Doppler tracking can result in complex LDV design and its performance can be limited by the vehicle velocity and acceleration.
A recently developed Laser Multi Beam Differential Interferometric Sensor (LAMBDIS) [12–13], provides measurement of vibration fields with interferometric sensitivity, while having low sensitivity to the motion of the sensor itself. The principle of operation of the LAMBDIS uses the same approach as in the parallel beam LDV, which measures velocity difference between two points of the object without employing a reference beam [14,15]. The LAMBDIS illuminates the object surface with a linear array of laser beams, and measures velocity difference between points on the object surface. Due to the absence of a reference beam LAMBDIS has low sensitivity to the motion of the sensor itself. Doppler shift induced by the sensor motion is canceled out that allows for measurements from a moving vehicle. Low sensitivity to the sensor motion allowed for the application of LAMBDIS for detection of buried objects from a moving vehicle. However, the use of an analog amplified photodiode array as a photodetector and a multichannel A/D converter in the LAMBDIS design increased the system dimensions and weight, making it difficult in practical applications [12].
In order to overcome these limitations, we developed a version of the LAMBDIS that employs a digital line-scan CMOS camera as a multichannel photodetector. To the best of our knowledge, this is the first use of a CMOS camera for a laser multi-beam differential interferometric vibration sensor. Recent progress in development of high-speed line-scan CMOS cameras allows for their application as photodetectors for multi-beam laser Doppler vibrometers [16]. Commercially available line-scan CMOS cameras have high frame rates (more than a hundred kilo-lines per second), high spatial resolution, and a large dynamic range. Optical signals on all pixels are processed simultaneously. Digital CMOS line-scan cameras combine thousands of photodiodes, amplifiers, and a digitizer in one compact package. Using CMOS line-scan cameras as photodetector for LAMBDIS allows for more compact design at a reduced cost.
Furthermore, the system was enhanced by incorporating an FPGA based real-time signal processor. Performance of the LAMBDIS based on a digital line-scan CMOS camera and FPGA signal processing for vibration imaging of buried objects was experimentally investigated. Description of the sensor and the experimental results are presented in the paper.
2. Functional layout and principle of operation
The functional layout of the LAMBDIS based on a line-scan CMOS camera is shown in Fig. 1. The principle of operation of the LAMBDIS is based on the interference of light reflected from different points on the object surface illuminated with a linear array of laser beams, as illustrated by Fig. 1. A linear array of 30 laser beams (only 6 beams are shown for clarity), is generated by the beam array generator of the sensor and focused onto an object (ground surface). Adjacent beams in the array have different optical frequencies, F1 and F2. Specifically, each beam of frequency F1 is positioned on the object surface exactly between two beams of frequency F2, so that the optical frequency of even number beams is different from optical frequency of odd number beams. A receiver lens and a shearing interferometer create two sheared images of the laser spots of the object on the sensor of the digital line-scan CMOS camera. The two images are sheared relative to each other in the direction of the array of points by an odd number of intervals between neighboring laser spots. In the image plane, the lights from each pair of corresponding laser spots are mixed together on the CMOS sensor, producing heterodyne signals with the carrier frequency ${F_C} = {F_2} - {F_1}$. For example, referring to Fig. 1, for the case of one interval shear between the two images, Image 1 and Image 2 of laser spots on the object surface, the Image 1 of spot 1 is overlapped with the sheared Image 2 of spot 2, the Image 1 of spot 2 is overlapped with the sheared Image 2 of the spot 3, and so on.
Fig. 1. Functional layout of LAMBDIS based on digital line-scan CMOS camera and FPGA-based real-time signal processor
Vibration of the object causes frequency shifts of reflected light due to the Doppler effect, resulting in frequency modulated signals with the carrier frequency ${F_C} = {F_2} - {F_1}$ on the CMOS camera. The vibration velocitiy difference between the corresponding points on the object, for example between points 1 and 2, points 2 and 3, and so on, are computed in real-time through demodulation of these heterodyne signals by a FPGA based processor. A vibration image of the measured area of the ground can be obtained by either scanning the array of beams over the area from a stationary platform using a scanner, or by moving the platform.
Motion of the sensor, or the whole-body motion of the object, causes practically the same Doppler shift for all beams, and is automatically subtracted from the measurements. So, the LAMBDIS principle allows for measuring vibration velocity between points on the object surface with interferometric sensitivity, while having low sensitivity to the motion of the sensor itself.
3. Optical schematic
The optical schematic of the LAMBDIS is shown in Fig. 2. The schematic works as follows.
Fig. 2. Optical schematic of LAMBDIS. M1-M8- mirror, NPBS1, 2 - non-polarizing beam splitter, AOM1, 2- acousto-optic modulator, BE- beam expander, DOE- diffraction optical element, L - lens.
The laser beam is divided by a non-polarizing beamsplitter NPBS1 into two beams, each of them is frequency shifted by a different amount of F1 = 110 MHz and F2 = 110.016 MHz respectively by using acousto-optical modulators AOM1 and AOM2. A dual-output AOM driver generates the two RF signals with F1 and F2 frequency respectively, derived from a common reference clock, making a highly stable carrier frequency ${F_C} = {F_2} - {F_1}$. Frequency shifted beams are then combined together on the focusable beam expander BE at an angle β to one another, using mirrors M1-M3. Frequency shifted beams pass through the beam expander, and are incident onto a diffractive optical element beam splitter DOE. The DOE splits frequency shifted beams into two identical linear arrays of beams. The beam expander increases the diameter of the beams and reduces the angle between the beams by M times, where M is the expansion ratio/magnification of the beam expander. The angle β between the beams incident on BE is related to the inter-beam angle Θ (the angle between neighboring beams at the DOE output) of the DOE and the magnification M of the beam expander through the following expression: $\beta = {{\Theta M} / 2}$. The DOE used in the design splits each incident beam into a line of 15 beams with inter-beam angle Θ = 0.407°, and full pattern angle 5.7°. The angle β between the two frequency shifted beams incident on the BE is adjusted in such a way that the two arrays of beams are sheared on the object surface relative to each other by a half of an inter-beam spacing of the array, producing a combined 30 beam linear array, in which adjacent beams have different optical frequencies F1 and F2. Specifically, the frequency of the neighboring beams in the combined beam array are 110 MHz and 110.016 MHz respectively, so each of the 110.0 MHz beams is positioned on the object exactly between two 110.016 MHz beams. Accordingly, in the linear array, there exists a 16 kHz frequency shift between neighboring beams. The interbeam angle of the combined 30 beam array is 0.2035°, and full pattern angle is 5.9°. The beams can be focused on the object surface by adjusting focus of the beam expander. The mirror M4 and the scanning mirror direct the laser beams to the object surface. The scanning mirror is used to scan the array of beams in a transverse direction across the object in order to create a vibration image of object surface. The light reflected back by the object surface is reflected by the scanning mirror, and mirrors M5 and M6, and enters a Michelson interferometer formed by non-polarizing beam splitting cube NPBS2 and mirrors M7 and M8. Laser radiations transmitted to and received from the object are incident on different portions of the scanning mirror. Such a configuration makes the transmitted and received radiations spatially separated on the scanning mirror in order to reduce direct coupling of transmitted light to the photodetector.
Lens L and the Michelson interferometer create two laterally sheared images of laser spots on the object surface on the sensor of a line-scan CMOS camera. As a result, the light from each pair of corresponding laser spots on the object, which have different frequencies are mixed together to interfere on the CMOS camera sensor, producing heterodyne signals with the carrier frequency $\; {F_C} = 16\; \textrm{kHz}$.
It should be noted that dynamic range of a CMOS camera is significantly lower than the dynamic range of photodiodes. In order to have interference signals within the dynamic range of the camera, the power of the laser beams should be selected accordingly, so that the intensity of image spots do not saturate the camera, and stay above the camera noise floor. The power of laser beams directed to the object can be adjusted by varying the laser output or controlling the RF signal amplitude of the AOM driver.
The light intensity $\; {I_{ij}}(t )\; $ of a heterodyne signal on the CMOS sensor can be written as
The first term in Eq. (2) ${F_{D0ij}}(t )= \frac{2}{\lambda }{v_0}({\cos {\alpha_{0i}} - \cos {\alpha_{0j}}} )$ represents the Doppler frequency of a LAMBDIS channel caused by the whole-body motion between the sensor and the object. For small angles (less than 1°) between two beams contributing to a heterodyne signal of a LAMBDIS channel, the Doppler shift ${F_{D0ij}}(t )$ caused by the sensor motion will be significantly lower than it would be for the Doppler shift in a traditional LDV for the same speed of motion. For example, for beams 1 and 2, with angles in the LAMBDIS design, as described below, ${\alpha _1}$= 2.85° and ${\alpha _2}$ = 2.65° relative to the optical axis, the Doppler shift caused by the sensor motion would be ${F_{D012}}(t )={-} 1.67\cdot {10^{ - 4}}\; \frac{2}{\lambda }{v_0}$, which is 1.67·10−4 of the Doppler shift in a LDV signal.
The second term in Eq. (2) ${F_{Dij}}(t )= \frac{2}{\lambda }({{v_i}\cos {\alpha_i} - {v_j}\cos {\alpha_j}} )$ represents the Doppler shift of a LAMBDIS channel caused by the vibration velocity difference of the object at points i and j. Using an FM-demodulation technique, the velocity difference ${v_i} - {v_j}$ between spots i and $j\; $ can be extracted, provided that angles ${\alpha _i}$ and ${\alpha _j}$ are known from the geometry of the sensor. For small, less than 1°, angles between two beams contributing to a heterodyne signal and the vibration velocity colinear with the direction of the laser beams, the expression for the Doppler shift caused by the vibration can be simplified to ${F_{Dij}}(t )= \frac{2}{\lambda }({{v_i} - {v_j}} )$ith an error of approximation less than 0.1%. One can conclude from the above estimation, that the LAMBDIS configuration provides measurements of the vibration velocity difference between points on the object surface with interferometric sensitivity and has low sensitivity to the motion of the sensor itself. Low Doppler shift caused by the sensor motion allows the sensor operation from a moving vehicle without employing Doppler tracking and increasing the processing bandwidth.
It should be noted that the Doppler shift of light reflected from each point on the object imparts adjacent channels with an opposite sign. As a result, the demodulated vibration velocity phase flips by 180 degrees for every other channel. The phase flip for every other channel can be explained considering the sign of Doppler shift of light reflected from neighboring points N, N + 1, and N + 2 on the object surface for the case when ${F_{DN}}(t )< {F_{DN + 1}}(t )< {F_{DN + 2}}(t ),$ where ${F_{DN}},$${F_{DN + 1}}$, and ${F_{DN + 2}}(t )$ are the Doppler shifts of light reflected from points N, N + 1, and N + 2 on the object respectively. The instantaneous Doppler frequency difference ${F_{DN + 1}}(t )- {F_{DN}}(t )$ between points N + 1 and N has the opposite sign relative to the instantaneous Doppler frequency difference ${F_{DN + 1}}(t )- {F_{DN + 2}}(t )$ between points N + 1 and N + 2. The origination of the phase flip can be also understood by considering the frequencies for two adjacent measurement channels: N and N + 1. Frequency ${F_N}(t )$ for channel N and frequency FN + 1 for channel N + 1 can be written as follows:
Signals from the line-scan CMOS camera are further processed in real time by a FPGA-based signal processing system to generate vibration images of the measured area.
4. Signal processing
The block-diagram of the signal processing system is shown in Fig. 3. The signal processing system consists of an FPGA board and a PC running MATLAB, see Fig. 3 (a). The FPGA can be logically partitioned into the Processing system (PS) - an embedded multi-core processor running a flavor of Linux and the Programmable logic (PL) – configurable and dedicated digital circuits that enable high-throughput signal processing.
Fig. 3. (a) The top-level block diagram of the signal processing system, (b) Block diagram of the section that computes the velocity magnitude and phase, BPF: Band-pass filter, LPF: Low-pass filter, FFT: Fast Fourier Transform.
A digital line-scan CMOS monochrome camera (Octoplus, Teledyne e2V) was used as a photodetector in the LAMBDIS design. The camera has the following major features: 2048 pixels line scan sensor, 10 × 200 µm pixel size, pixel full well capacity 200ke, USB 3 interface, bit depth 12 bits, and the maximum line rate of 120 kl/s.
The line-scan camera interacts with the FPGA PS via a USB 3 driver, that supports a continuous data rate of 370 MB/s. The PS handles the data moving operations from the camera to the PL and from the PL to the PC. The PL performs the pixel arrangement, vibration velocity magnitude and phase calculation, and also interfaces with the audio card for the audio reference input. This audio reference signal is the signal used for vibration excitation and is used to obtain the vibration velocity phase during scanning. Figure 3 (b) shows the method used for calculation of the vibration velocity magnitude and phase. The velocity is obtained by using frequency demodulation of heterodyne signals using an I&Q demodulation technique [4,12]. The real-valued amplitude data coming from the line-scan camera is converted to complex-valued data using the Hilbert transform. A band-pass filter is employed to select the spectral region of interest and to reject noise. The carrier frequency value defined by a user though the graphical user interface (GUI) is generated in the Carrier generator block, which generates the local oscillator (LO) signal. The LO and the complex-valued data from the
BPF are mixed and low-pass filtered to effectively demodulate the vibration velocity information from the FM carrier signal. The argument i.e. phase of the complex-valued demodulated data is then computed using the CORDIC (Coordinate rotation digital computer) method [17]. Subsequently, the phase difference between successive values of demodulated data, which is proportional to the vibration velocity at that instant, is computed. At this point in the data flow, the proportional vibration velocity data are available at the sampling rate of the line-scan camera. In order to work with the limited RAM available on the FPGA, the velocity spectrum, and consequently the sampling rate, is progressively trimmed as the data moves downstream, to remove frequencies outside the region of interest. In the time domain, a lower sampling rate allows for longer contiguous blocks of data to be held in the available FPGA RAM. An FFT of the proportional vibration velocity data over longer time-domain data blocks gives higher spectral resolution for the vibration velocity. From this complex-valued velocity spectrum, the velocity magnitude and phase for the user selected frequency bins are computed. In the Phase correction block, the velocity phase computed is corrected using the phase information obtained from the audio reference. The magnitude and phase information is sent to the PC, where the MATLAB code scales the velocity magnitude to display the absolute (not proportional) vibration velocity images on the user interface. This interface allows the user to set various system parameters, including the control of the mirror scanner.
Pixel selection takes place on the PC. The optical signal received by the line-scan camera consists of a series of peaks corresponding to the images of laser spots on the object surface. An example of an output signal of the line-scan camera is shown in Fig. 4. The output signal consists of 33 peaks, corresponding to a combined image of two arrays of 30 spots sheared by three intervals between the spots. In the combined image 27 spots are overlapped producing heterodyne signals resulting in 27 measurement channels. Three peaks on the left and right sides of the combined image don’t produce heterodyne signals and are excluded from processing by a user in the GUI. The 2048 pixels of the line-scan camera are arranged in 512 groups of 4 pixels each, which we refer to as a quad pixel. Quad pixels with amplitudes that lie between a low and high thresholds selected by a user are selected as measurement channels. Selected low and high thresholds depend on object reflectance, laser beams power, and distance to the object, and can vary for different situations. The velocity phase of adjacent measurement channels differs by π radians, as shown in expressions (3) and (4). This information is passed from the pixel selection method on the PC to the FPGA before each data capture. The FPGA uses this information to correct for the phase of adjacent measurement channels, resulting in proper display of velocity phase and instantaneous velocity images on the GUI. The FPGA computes the velocity amplitude and phase of all 512 quad pixels and sends this information to the PC. The user interface on the PC plots vibration images using only those quad pixels that are selected as measurement channels by the pixel selection method. The bulk of the data processing takes place in real-time on the FPGA. The PC MATLAB code primarily handles the user interface and spectrogram display, which are not computationally intensive and thus we are able to maintain the real-time capability for the entire (FPGA + PC) system.
Fig. 4. Output signal on the line-scan camera corresponding to the image of the linear array of beams.
5. Experimental results
Performance of the LAMBDIS based on a line-scan CMOS camera and FPGA signal processing was experimentally verified in laboratory by conducting real-time vibration imaging of a circumferentially clamped 250 mm diameter plate. The LAMBDIS measures vibration velocity difference between points on the object and shows a spatial differential velocity profile which could be treated as a deformation gradient of the object surface [12]. The plate vibrating in a first axial-symmetric spatial mode at 314 Hz natural frequency was measured by scanning the LAMBDIS beams across the object at a speed of beams of 20 cm/s. The two images of the laser spots on the CMOS camera were sheared relative to each other by three intervals between neighboring spots in the image plane. As a result, there were 27 measurement channels producing heterodyne signals. Figure 5 shows an example of a real-time differential velocity image for the velocity magnitude (a), instantaneous velocity (b), and vibration phase (c). Instantaneous velocity and phase plots were obtained for the extreme position of the object in the vibration cycle.
Fig. 5. Vibration velocity image of a circumferentially clamped circular plate obtained with the LAMBDIS: (a) – differential velocity magnitude, (b) – differential instantaneous velocity, (c) -vibration phase.
The plate was also measured for comparison purposes with a traditional single-beam scanning LDV, PSV 500 (Polytec, Inc.). The velocity magnitude (a), instantaneous vibration velocity (b), and vibration phase (c) of the plate are shown in Fig. 6.
Fig. 6. Vibration velocity image of a circumferentially clamped circular plate obtained with a scanning LDV: (a) - velocity magnitude, (b) – instantaneous velocity, (c) -vibration phase.
One can see from comparison of Figs. 5 and 6, that LAMBDIS measures deformation gradient, while a traditional LDV measures deformation of the object surface. The vibration velocity difference between two points of the object surface measured with the LAMBDIS was compared with the velocity difference between the same points measured with a single beam LDV (PDV 100, Polytec, Inc.). The measurement results of the velocity difference vs. the velocity at one of the measured points are presented in Fig. 7, and show a very good agreement between the LAMBDIS and LDV measurements.
The ability of the LAMBDIS to produce real-time vibration images of low amplitude vibrations was verified by measuring the vibrating plate for different vibration amplitudes. Figure 8 shows and example of instantaneous differential vibration velocity images of the vibrating plate obtained for the extreme position of the object in the vibration cycle for different vibration amplitudes at the center of the plate: 160 µm/s - (a), 83 µm/s - (b), and 19 µm/s - (c).
Fig. 8. Instantaneous differential vibration velocity images of the vibrating plate for different vibration amplitude at the center of the plate: 160 µm/s - (a), 83 µm/s - (b), and 19 µm/s - (c).
Figure 9 shows the spatially filtered instantaneous vibration velocity images of the vibrating plate shown in Fig. 8.
Fig. 9. Spatially filtered instantaneous differential vibration velocity images of the vibrating plate for different vibration amplitude at the center of the plate: 160 µm/s - (a), 83 µm/s - (b), and 19 µm/s - (c).
One can see from Figs. 8 and 9 that the LAMBDIS is capable of measuring vibrations with differential velocities below 2 µm/s.
Field experiments on detection of buried objects have been conducted with the LAMBDIS mounted on a small electric vehicle shown in Fig. 10. The LAMBDIS optical head, and electronics including the FPGA unit, the PC, and the AOM driver are mounted on the front of the vehicle and a power supply is mounted in a 19-inch rack on the back of the vehicle. A lifting mechanism allows adjustment of the height of the optical head from 1.50 m to 2.2 m measured at the scanning mirror.
The geometrical configuration of the setup for field experiments consists of a linear array of 30 beams from the LAMBDIS. The height of beams above the ground was 2.2 m measured at the scanning mirror, and the standoff distance was 8 m resulting in a beam angle of 15.4° with the ground surface. Given the full angle of the beam pattern of Θ = 5.9° and the inter-beam angle of 0.203°, the spread of beams D on the ground was D = 82 cm and the distance between the neighboring beams was d = 2.8 cm at 8 m standoff distance. For the field experiments, the LAMBDIS alignment was made in such a way that two images of the laser spots on the CMOS camera were sheared relative each other by three intervals between neighboring laser spots in the image plane. As a result, the LAMBDIS measured differential velocities between points of the object separated by three intervals between the points: channel 1 provided differential velocity between points 1 and 4, channel 2 - between points 2 and 5, and so on. With 30 beams on the ground, there were 27 measurement channels.
A vibration image of the ground area was obtained by scanning the ground with LAMBDIS beams in two modes of operation:
- 1. Moving mode. Beams scan the ground due to the vehicle motion. The sensor is moving forward and the scanning mirror is stationary. The scanning mirror angle defines the standoff distance.
- 2. Scanning mode. The sensor/vehicle is stationary. Initial angle of the scanning mirror defines the standoff distance. The scanning mirror scans the beams along the track by a predetermined distance, for example one meter. The speed of scanning and the scanning distance can be selected by the operator.
Figure 12 shows an example of real-time vibration images of one square meter of the ground surface above a buried object obtained in a scanning mode. Figure 12 (a) shows a differential velocity magnitude image, and Fig. 12 (b) shows a differential instantaneous velocity image obtained for the extreme position of the object in the vibration cycle. Figure 13 shows an example of real-time vibration images of one square meter of the ground surface above a buried object obtained in a moving mode. Figure 13 (a) shows a differential velocity magnitude image, and Fig. 13 (b) shows a differential instantaneous velocity image obtained for the extreme position of the object in the vibration cycle.
Fig. 12. Vibration image of a buried object obtained in real time from a stationary vehicle using the scanning mirror: (a) – differential velocity magnitude, (b) – differential instantaneous velocity
Fig. 13. Vibration image of a buried object obtained in real time from a moving vehicle: (a) – differential velocity magnitude, (b) – differential instantaneous velocity
The Y-axis in Figs. 12 and 13 shows the time of scanning the ground area over one meter, the X-axis shows the number of measurement channels. The scanned area was 1 × 1 meter, and the speed of beams was 10 cm/s for both measurements from a stationary and a moving vehicle. The field experiments confirm that the LAMBDIS based on a line-scan CMOS camera and FPGA signal processing is able to detect, in real time, buried objects from a stationary and a moving vehicle.
6. Conclusion
A novel Laser Multi Beam Differential Interferometric Sensor (LAMBDIS) that employs a digital line-scan CMOS camera and FPGA based real-time signal processing has been developed. The LAMBDIS is able to measure differential vibration velocities with interferometric sensitivity between points on the object illuminated with a linear array of thirty laser beams, while having low sensitivity to the motion of the sensor itself. A vibration image of an object can be obtained by scanning a linear array of beams over the object or by moving the sensor platform. Low sensitivity to the motion of the sensor itself allows using the LAMBDIS for vibration measurements from a moving platform. The LAMBDIS was successfully used for real-time vibration imaging of the ground surface in field experiments on laser-acoustic detection of buried objects. The field experiments demonstrated the ability of LAMBDIS to detect buried objects from a moving vehicle in real time.
Funding
Office of Naval Research (N00014-18-1-2489).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. J.M. Sabatier and N. Xiang, “An Investigation of Acoustic-to-Seismic Coupling to Detect Buried Antitank Landmines,” IEEE Trans. on Geosci. Remote Sens. 39(6), 1146–1154 (2001). [CrossRef]
2. N. Xiang and J.M. Sabatier, “An Experimental Study on Antipersonnel Landmine Detection using Acoustic-to-Seismic Coupling,” J. Acoust. Soc. Am. 113(3), 1333–1341 (2003). [CrossRef]
3. N. Xiang and J.M. Sabatier, “Laser Doppler Vibrometer-Based Acoustic Landmine Detection Using the Fast M-Sequence Transform,” IEEE Geosci. Remote Sensing Lett. 1(4), 292–294 (2004). [CrossRef]
4. V. Aranchuk, A. Lal, C. Hess, and J.M. Sabatier, “Multi-beam laser Doppler vibrometer for landmine detection,” Opt. Eng. 45(10), 104302 (2006). [CrossRef]
5. A. Lal, V. Aranchuk, V. Doushkina, E. Hurtado, C. Hess, J. Kilpatrick, D. L’Esperance, N. Luo, V. Markov, J. Sabatier, and E. Scott, “Advanced LDV instruments for buried landmine detection,” Proc. SPIE 6217, 621715 (2006). [CrossRef]
6. V. Aranchuk, A. Lal, H. Zhang, C. Hess, and J.M. Sabatier, “Acoustic Sensor for Landmine Detection using a Continuously Scanning Multi-Beam LDV,” Proc. SPIE 5415, 61–69 (2004). [CrossRef]
7. V. Aranchuk, J.M. Sabatier, A.K. Lal, C.F. Hess, R.D. Burgett, and M. O’Neill, “Multi-beam laser Doppler vibrometry for acoustic landmine detection using airborne and mechanically-coupled vibration,” Proc. SPIE 5794, 624–631 (2005). [CrossRef]
8. A.K. Lal, C.F. Hess, H. Zhang, E. Hurtado, V. Aranchuk, V.B. Markov, and W. T Mayo, “Whole-field laser vibrometer for buried landmine detection,” Proc. SPIE 4742, 640–648 (2002). [CrossRef]
9. H. Kim, Y. Lee, C. Kim, T.-G. Chang, and M.-S Kang, “Laser Doppler vibrometer with body vibration compensation,” Opt. Eng. 42(8), 2291–2295 (2003). [CrossRef]
10. L. A. Jiang, M.A. Albota, R.W. Haupt, J.G. Chen, and R.M. Marino, “Laser vibrometry from a moving ground vehicle,” Appl. Opt. 50(15), 2263–2273 (2011). [CrossRef]
11. B.J Halkon and S. J Rothberg, “Taking laser Doppler vibrometry off the tripod: correction of measurements affected by instrument vibration,” Opt. Lasers Eng. 91, 16–23 (2017). [CrossRef]
12. V. Aranchuk, I. Aranchuk, B. Carpenter, C. Hickey, D. Kleinert, H. Buchanan, and J.D. Heffington, “Laser multi-beam differential interferometric sensor for acoustic detection of buried objects,” Proc. SPIE 10628, 1062804 (2018). [CrossRef]
13. V. Aranchuk, I. Aranchuk, B. Carpenter, and C. Hickey, “Laser Doppler multi-beam differential vibration sensor for acoustic detection of buried objects,” in Laser Congress 2019 (ASSL, LAC, LS&C), OSA Technical Digest (Optical Society of America, 2019), paper LM4B.2.
14. N.A. Halliwell, C.J.D. Pickering, and P.G. Eastwood, “The laser torsional vibrometer: a new instrument,” J. Sound Vibrat. 93(4), 588–592 (1984). [CrossRef]
15. J.R. Bell and S.J. Rothberg, “Rotational vibration measurements using laser Doppler vibrometry: comprehensive theory and practical application,” J. Sound Vibrat. 238(4), 673–690 (2000). [CrossRef]
16. V. Aranchuk, R. Kasu, J. Li, I. Aranchuk, and C. Hickey, “Multi-beam heterodyne laser Doppler vibrometer based on a line-scan CMOS digital camera,” Appl. Opt. 61(20), 5876–5883 (2022). [CrossRef]
17. CORDIC (xilinx.com)
