A high-precision calibration method for a hyperspectral lidar for photon counting

Abstract

The application discloses a high-precision calibration method for a hyperspectral laser radar for photon counting, belongs to the technical field of laser radars, and comprises two stages of spectral calibration and radiation calibration; the spectral calibration is based on monochromator wavelength-by-wavelength scanning and synchronous measurement of a beam splitter to extract spectral response parameters of each channel; the radiation calibration strictly decouples a detection link, extracts an effective attenuation coefficient of water in situ by using a slope method, and combines an air white plate benchmark and a window transmittance to perform full-link radiation correction, and inversely calculates target intrinsic spectral reflectivity which is stripped of interference. The application can be used for data correction, target accurate classification and identification and high-fidelity quantitative detection of the hyperspectral laser radar for photon counting, and solves the problem that original signals produce serious spectral distortion due to non-uniform energy of a light source of the hyperspectral laser radar and complex scattering and absorption of water.

Description

Technical Field

[0001] This invention belongs to the field of lidar technology, and in particular relates to a high-precision calibration method for hyperspectral lidar used for photon counting. Background Technology

[0002] Hyperspectral lidar, as an emerging active remote sensing technology, integrates the advantages of ranging and multi-band imaging. It can simultaneously acquire the "spectral" information of a target by emitting supercontinuum lasers and using a multi-channel receiving architecture. This technology has great potential for fine classification of complex scenes, but faces two core obstacles in practical engineering implementation. Firstly, the system must efficiently and losslessly split and couple the echo signals across an extremely wide spectral band to maintain overall operational stability. Secondly, the uneven energy distribution across different bands of the supercontinuum light source results in weak echo signals in some bands, increasing the difficulty of accurate photon signal detection and system calibration.

[0003] Extracting the true intrinsic spectral reflectance of the target is the core task of this detection system.

[0004] Chinese patent document CN120541367A discloses an online processing system and method for multi-channel echo data of lidar. The online processing system includes a data distribution layer, which segments the echo data and sends it to a data buffer layer to extract local maxima, the corresponding filtered sampling point subsequences, and the index of the local maxima within the corresponding filtered sampling point subsequences. These are then input as pre-fitting data to the data processing layer for pre-fitting and fitting based on a skewed normal waveform fitting function, resulting in optimized signal strength and optimized waveform center position, thereby further obtaining the target's channel calculation distance. The online processing method utilizes the online processing system to obtain the target's spatial location information, corrected spectral information, and spectral reflectance. However, this method mainly relies on pure mathematical fitting and local feature extraction of the echo waveform, failing to address the core challenge of extremely uneven energy distribution across a wide spectral range in supercontinuum light sources at the physical level. Furthermore, the system lacks an absolute radiometric calibration mechanism for attenuation in complex transmission media (such as water) and differences in the multi-channel response of the system, making it difficult to accurately deduce the intrinsic spectral reflectance of the target under complex detection conditions with severe spectral distortion.

[0005] Chinese patent document CN117665744A discloses a method for processing echo waveforms of highly overlapping hyperspectral lidar. Based on finite element method calculations of the laser spot area, the method integrates the echo waveform of each surface element within the spot to obtain the final waveform simulation result. For multi-channel hyperspectral lidar echo waveforms, it confirms the presence of multiple targets in the echo based on range resolution enhancement. Due to the different spectral reflectivities of different targets, the centroid of the multi-channel measurement echo waveform of the hyperspectral lidar shifts towards the target with higher spectral reflectivity or lower tilt. By obtaining the waveform less affected by spectral or tilt factors and subtracting it from the waveform most affected by these factors, the difference waveform provides effective clues to the number and spatial location information of waveforms within the echo waveform. However, this method focuses on spatial location analysis and waveform difference calculation under multi-target waveform superposition, and essentially still falls within the scope of relative signal processing, failing to establish a complete end-to-end radiometric calibration correction model. This method cannot completely eliminate systematic errors such as optical response deviation, window transmittance loss, and two-way attenuation in the water medium within the instrument. Therefore, it cannot be directly applied to the quantitative physical inversion of the absolute spectral reflectance of unknown underwater targets.

[0006] Faced with the non-uniformity of energy from white light sources and the complex scattering and absorption of beams of different wavelengths by the transmission medium, the raw signal captured by the receiver will suffer severe spectral distortion. To achieve accurate target identification using pure reflectivity data during the point cloud processing stage, the system must rely on a rigorous calibration model to completely eliminate internal response biases and external medium transmission losses. Therefore, developing a customized high-precision calibration method tailored to the optomechanical characteristics of photon-counting hyperspectral lidar becomes a crucial step in improving its detection performance and data availability. Summary of the Invention

[0007] To address the precision calibration problem of photon-counting hyperspectral lidar systems, this invention provides a high-precision calibration method for photon-counting hyperspectral lidar, which can efficiently achieve dual calibration of the system's spectral and radiation characteristics, providing key technical support for establishing physical benchmarks for various multi-channel broadband imaging radars in the field of remote sensing.

[0008] A high-precision calibration method for hyperspectral lidar used for photon counting includes:

[0009] Spectral calibration stage:

[0010] S1a: Connect the supercontinuum laser to the monochromator, set the wavelength scanning range and scanning step size for scanning; connect the beam splitter at the output end of the monochromator to split the light into two paths, with 10% of the energy entering the reference light path and 90% of the energy entering the main detector light path.

[0011] S2a: The net photon count of the reference optical path and the main detector optical path are obtained respectively, and the response ratio is calculated. Then, the center wavelength and effective spectral bandwidth of each independent detector channel are extracted.

[0012] Radiation calibration phase:

[0013] S1b collects the dark count when the laser is off, the ambient background noise when there is no target, and the original aliasing signal during normal detection, and extracts the net photon signal by stripping it off; the net photon signal is divided into a window reflection window, a water body scattering window, and a target echo window on the time axis.

[0014] S2b, measure the standard whiteboard in the air, integrate the net photon count in the target echo window to obtain the air whiteboard reference signal;

[0015] S3b records the reference signals and corresponding measurement distances when the whiteboard is directly viewed from the air and when the whiteboard is detected through the window glass, respectively, and calibrates the two-way transmittance of the window.

[0016] S4b extracts echo data within the water body scattering window, calculates the effective attenuation coefficient of the water body in each band, and obtains the two-way transmittance of the water body.

[0017] S5b performs time window integration on the echo signal of an unknown underwater target to obtain the total target signal, and uses the whiteboard reference signal data to calculate the initial reflectivity of the target when water attenuation is ignored.

[0018] S6b substitutes the initial reflectance of the target into the full-link radiative transfer correction model, removes the two-way transmittance of the viewing window and the two-way transmittance of the water body, and inverts the intrinsic spectral reflectance of the underwater target.

[0019] In step S1a, the wavelength scanning range is set to 460nm to 800nm ​​and a scanning step size of 1nm is maintained.

[0020] In step S2a, the formula for calculating the response ratio is:

[0021] ;

[0022] in, The ratio of the response of the main probe optical path to that of the reference optical path. The net photon count of the main detector optical path, For reference, net photon count of the optical path, This indicates the center wavelength of the corresponding detection channel.

[0023] In step S2a, the center wavelength and effective spectral bandwidth of each independent detection channel are extracted. The specific process is as follows:

[0024] The response ratio sequences of 32 independent detection channels were fitted with Gaussian functions. The center wavelength was extracted from the peak value of the fitted curve, and the full width at half maximum (FWHM) was determined as the effective spectral bandwidth of each channel.

[0025] In step S2b, a standard white board is placed in the air, and the raw photon events of each detection channel are collected synchronously. Dark counts and environmental background noise are strictly subtracted, and the net photon count within the time window is calculated. Integrate to obtain an air whiteboard reference signal that eliminates internal system fluctuations. :

[0026] ;

[0027] in, Indicates the target echo window.

[0028] In step S3b, the formula for calibrating the two-way transmittance of the viewing window is:

[0029] ;

[0030] in, Two-way transmittance of the window As a reference signal that penetrates the viewing window glass, As a reference signal for direct airflow, This indicates the sequence number of the independent detection channel (the value range corresponds to the 32 channels of the system). and These are the corresponding measured distances.

[0031] In S4b, the effective attenuation coefficient of water in each band is calculated. The specific process is as follows:

[0032] Echo data within the water body scattering window are extracted, binned and accumulated according to depth coordinates, and distance spherical divergence geometric compensation is performed. Then, a local homogeneous water body interval is selected in the logarithmic domain for linear fitting, and the effective attenuation coefficient of the water body in each band is calculated in situ.

[0033] By performing linear fitting on a locally homogeneous water body within the logarithmic domain, the following fitted model was obtained:

[0034] ;

[0035] in, The effective attenuation coefficient of the water body. , and These are the target reflectance and the standard white board reflectance, respectively. Set the calibration distance for the whiteboard; The corrected strength after geometric compensation and satisfies , For depth Volume scattering intensity at that location This indicates the reference signal for the air whiteboard.

[0036] The formula for obtaining the two-way permeability of water is as follows:

[0037] ;

[0038] in, Two-way water permeability The effective attenuation coefficient of the water body. This represents the actual geometric depth of laser transmission in the water.

[0039] In step S5b, the formula for calculating the initial reflectivity of the target is:

[0040] ;

[0041] in, For the target echo total signal, This serves as the reference signal for the air whiteboard. For underwater target detection range, To calibrate the distance to the whiteboard, The reflectance is that of a standard whiteboard.

[0042] In step S6b, the intrinsic spectral reflectance of the underwater target is retrieved using the following formula:

[0043] ;

[0044] in, The initial reflectivity of the target. For window two-way transmittance, This represents the two-way permeability of water.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] This invention, through a two-stage "spectral-radiative" calibration and a full-link interference correction mechanism, not only achieves channel-level precision calibration over a wide spectral range but also creatively introduces a logarithmic domain slope method to extract the attenuation characteristics of the transmission medium. This mechanism enables absolute radiometric self-calibration of the system without relying on external measuring instruments, significantly reducing error coupling interference caused by complex detection environments. It provides an extremely reliable technical guarantee for photon-counting hyperspectral lidar to achieve high-fidelity quantitative detection with "image-spectrum integration." Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a schematic diagram of a high-precision calibration method for a hyperspectral lidar used for photon counting, according to an embodiment of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] It should be noted that, unless otherwise specified, the features in the following embodiments and implementation methods can be combined with each other.

[0051] like Figure 1 As shown, a high-precision calibration method for hyperspectral lidar used for photon counting includes two stages: spectral calibration and radiometric calibration.

[0052] The spectral calibration stage is based on the core architecture of "monochromatic wavelength-by-wavelength scanning + beam splitter synchronous measurement". A calibration platform is built in a controlled laboratory environment to accurately determine the center wavelength, spectral bandwidth, and spectral response range of the 32 detection channels of the photon-counting hyperspectral lidar. The specific implementation steps are as follows:

[0053] The system's built-in supercontinuum laser is connected to a high-precision monochromator via fiber optic cable. The monochromator's output end is connected to a 10:90 beam splitter, splitting the light into two paths. 10% of the energy enters the reference optical path and directly connects to the reference photon counter, while the remaining 90% of the energy passes through the radar system components and then enters the 32-channel system photon counter. The monochromator's scanning range is set to 460nm~800nm ​​with a fixed 1nm step size. The beam splitter's incident end is completely blocked with a light shield, and 100 sets of data are collected and averaged to complete the dark count calibration of the two photon counters, obtaining the dark count baselines for the reference optical path and the system optical path, respectively.

[0054] Remove the light shield and start the supercontinuum laser to preheat until the power is stable. The monochromator performs a full-band scan from 460nm to 800nm ​​in 1nm steps. At each wavelength node, two counters are triggered synchronously to collect the total number of photons in the reference optical path and the system optical path and strictly subtract the dark count baseline. To ensure data stability, three sets of data are collected every 10nm and the average value is calculated.

[0055] Calculate the relative response ratio of the net photon count recorded in the main detector path to the net photon count in the reference path:

[0056] ;

[0057] The data sequences corresponding to the 32 channels were extracted and fitted with a Gaussian function. The peak coordinates of the fitted curve were used as the center wavelength of the corresponding channel. And according to the formula • σ is used to calculate the effective spectral bandwidth, and then to determine the spectral response range of each channel. .

[0058] Following spectral calibration, further full-link absolute radiometric calibration is required. The core theory lies in strictly decoupling the radar detection link into "system term × glass term × water term × target reflectance term." By calibrating layer by layer and stripping away the attenuation of each medium, the intrinsic spectral reflectance of the underwater target can be accurately inverted. The specific implementation steps are as follows:

[0059] The data acquisition device performed dark counting under three operating conditions: laser obstruction off, laser on but no target, and normal detection and measurement. Environmental background noise and the original aliasing signal The net photon signal was obtained by subtraction stripping. For the net photon sequence of each detection channel, three non-overlapping characteristic time windows are precisely divided on the time axis: window reflection window. Water body scattering window and target echo window .

[0060] Set up standard whiteboards in the air and set the target echo time window. By integrating the net photon number within the system, an air whiteboard reference signal is obtained to eliminate energy fluctuations within the system. .

[0061] To eliminate the optical loss of the watertight chamber, the reference signals and corresponding measurement distances were recorded when detecting the white board without a glass window (direct view of the white board through the air) and when the white board was detected through the viewing window glass. This allowed for the absolute determination of the two-way transmittance of the viewing window glass.

[0062] ;

[0063] The position 20% of the leading edge of the reflection peak extracted from the glass window is defined as the zero point of laser water entry time. Subsequently, multiple sets of known true distances were found below the water surface. Place a flat target at the location and record the corresponding echo peak time. Construct and fit the distance curve in water ,in The equivalent speed of light scaling factor is obtained from linear fitting. The system zero-point residual deviation obtained from the fitting can then be used to achieve a precise mapping from photon flight time to underwater geometric depth.

[0064] ;

[0065] Extracting water body volume scattering window The volume scattering intensity is obtained by binning and accumulating the echo data within the depth coordinate. Using air reference signals Normalize the system response and multiply by the depth squared. To compensate for the geometric loss caused by beam divergence, the output correction intensity is:

[0066] ;

[0067] Within the logarithmic domain, a locally homogeneous water body is selected, and a linear equation is constructed based on the slope method:

[0068] ;

[0069] in, , and The reflectances of the target and the whiteboard are respectively; the slope of the straight line is extracted by least squares fitting, and the effective water attenuation coefficient of each band is calculated in situ. And based on the Lambert-Beer law, it is converted into the one-way transmittance at any depth in water:

[0070] ;

[0071] The total signal of the target is obtained by integrating the echo signal of an unknown underwater target within a time window. By using baseline data of an air whiteboard with known reflectivity to eliminate light source fluctuations and detector gain differences, the initial reflectivity of the target, neglecting water attenuation, was calculated:

[0072] ;

[0073] Substituting the initial reflectance into the end-link radiative transfer correction model, and strictly eliminating the previously independently calibrated window two-way transmittance and the water body two-way transmittance derived from the effective attenuation coefficient, the intrinsic spectral reflectance of the underwater target completely stripped of interference is retrieved:

[0074] .

[0075] The embodiments described above provide a detailed explanation of the technical solutions and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, additions, and equivalent substitutions made within the scope of the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A high-precision calibration method for hyperspectral lidar used for photon counting, characterized in that, include: Spectral calibration stage: S1a, connect the supercontinuum laser to the monochromator, set the wavelength scanning range and scanning step size for scanning; The monochromator output end is connected to a beam splitter to split the light into two paths: 10% of the energy enters the reference light path, and 90% of the energy enters the main detector light path. S2a: The net photon count of the reference optical path and the main detector optical path are obtained respectively, and the response ratio is calculated. Then, the center wavelength and effective spectral bandwidth of each independent detector channel are extracted. Radiation calibration phase: S1b collects the dark count when the laser is off, the ambient background noise when there is no target, and the original aliasing signal during normal detection, and extracts the net photon signal by stripping it off; the net photon signal is divided into a window reflection window, a water body scattering window, and a target echo window on the time axis. S2b, measure the standard whiteboard in the air, integrate the net photon count in the target echo window to obtain the air whiteboard reference signal; S3b records the reference signals and corresponding measurement distances when the whiteboard is directly viewed from the air and when the whiteboard is detected through the window glass, respectively, and calibrates the two-way transmittance of the window. S4b extracts echo data within the water body scattering window, calculates the effective attenuation coefficient of the water body in each band, and obtains the two-way transmittance of the water body. S5b performs time window integration on the echo signal of an unknown underwater target to obtain the total target signal, and uses the whiteboard reference signal data to calculate the initial reflectivity of the target when water attenuation is ignored. S6b substitutes the initial reflectance of the target into the full-link radiative transfer correction model, removes the two-way transmittance of the viewing window and the two-way transmittance of the water body, and inverts the intrinsic spectral reflectance of the underwater target.

2. The high-precision calibration method for hyperspectral lidar used for photon counting according to claim 1, characterized in that, In step S1a, the wavelength scanning range is set to 460nm to 800nm, and a scanning step size of 1nm is maintained.

3. The high-precision calibration method for hyperspectral lidar used for photon counting according to claim 1, characterized in that, In step S2a, the formula for calculating the response ratio is: ； in, The ratio of the response of the main probe optical path to that of the reference optical path. The net photon count of the main detector optical path, For reference, net photon count of the optical path, This indicates the center wavelength of the corresponding detection channel.

4. The high-precision calibration method for hyperspectral lidar used for photon counting according to claim 1, characterized in that, In step S2a, the center wavelength and effective spectral bandwidth of each independent detection channel are extracted. The specific process is as follows: The response ratio sequences of 32 independent detection channels were fitted with Gaussian functions. The center wavelength was extracted from the peak value of the fitted curve, and the full width at half maximum (FWHM) was determined as the effective spectral bandwidth of each channel.

5. The high-precision calibration method for hyperspectral lidar used for photon counting according to claim 1, characterized in that, In step S3b, the formula for calibrating the two-way transmittance of the viewing window is: ； in, Two-way transmittance of the window As a reference signal that penetrates the viewing window glass, As a reference signal for direct airflow, Indicates the serial number of the independent detection channel. and These are the corresponding measured distances.

6. The high-precision calibration method for hyperspectral lidar used for photon counting according to claim 1, characterized in that, In S4b, the effective attenuation coefficient of water in each band is calculated. The specific process is as follows: Echo data within the water body scattering window are extracted, binned and accumulated according to depth coordinates, and distance spherical divergence geometric compensation is performed. Then, a local homogeneous water body interval is selected in the logarithmic domain for linear fitting, and the effective attenuation coefficient of the water body in each band is calculated in situ.

7. The high-precision calibration method for hyperspectral lidar for photon counting according to claim 6, characterized in that, By performing linear fitting on a locally homogeneous water body within the logarithmic domain, the following fitted model was obtained: ； in, The effective attenuation coefficient of the water body. , and These are the target reflectance and the standard white board reflectance, respectively. Set the calibration distance for the whiteboard; The corrected strength after geometric compensation and satisfies , For depth Volume scattering intensity at that location This indicates the reference signal for the air whiteboard.

8. The high-precision calibration method for hyperspectral lidar for photon counting according to claim 6, characterized in that, The formula for obtaining the two-way permeability of water is as follows: ； in, Two-way water permeability The effective attenuation coefficient of the water body. This represents the actual geometric depth of laser transmission in the water.

9. The high-precision calibration method for hyperspectral lidar used for photon counting according to claim 1, characterized in that, In step S5b, the formula for calculating the initial reflectivity of the target is: ； in, For the target echo total signal, This serves as the reference signal for the air whiteboard. For underwater target detection range, To calibrate the distance to the whiteboard, The reflectance is that of a standard whiteboard.

10. The high-precision calibration method for hyperspectral lidar for photon counting according to claim 1, characterized in that, In step S6b, the intrinsic spectral reflectance of the underwater target is retrieved using the following formula: ； in, The initial reflectivity of the target. For window two-way transmittance, This represents the two-way permeability of water.

Images