Unmanned aerial vehicle sensing method and system based on laser and multi-band infrared cooperation

Abstract

The application provides a UAV sensing method and system based on laser and multi-band infrared cooperation, and relates to the technical field of UAV environment sensing. The application collects atmospheric scattering signals and infrared radiation data through laser radar and multi-band infrared sensors; inverts atmospheric attenuation coefficients and particulate matter distribution information as atmospheric state data, and analyzes background heat distribution as background data; generates a phase mask according to the atmospheric state data to modulate laser echo and infrared radiation wavefront coding to obtain an intermediate image; reconstructs an initial laser radar intensity image and an initial infrared thermal image with the atmospheric state data and the background data as prior constraints; jointly processes to separate common attenuation components and independent target radiation components; inputs the target image into a cooperative sensing model, analyzes polarization differences, and jointly interprets material, edge contour and thermal radiation information, separates obstacles, terrain features and passable areas, can effectively suppress atmospheric interference, and improves the sensing ability of the UAV under complex weather conditions.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) environmental perception technology, and in particular to a UAV perception method and system based on the collaboration of laser and multi-band infrared. Background Technology

[0002] In various applications such as military reconnaissance, disaster search and rescue, power line inspection, precision agriculture, and autonomous logistics delivery, accurate perception of the surrounding environment is a core prerequisite for safe flight and autonomous decision-making. As the complexity of drone missions continues to increase, traditional perception solutions relying on a single sensor mode are increasingly revealing their limitations, especially under complex weather conditions such as fog, dust, rain, and snow, where the perception performance of a single-mode sensor will significantly degrade.

[0003] LiDAR, as an active ranging sensor, provides high-precision three-dimensional spatial information and target reflectance intensity data, offering significant advantages in target detection and distance measurement. However, laser signals are highly susceptible to aerosol scattering and molecular absorption during atmospheric transmission, leading to severe echo signal attenuation, reduced detection range, and decreased target recognition rate. Multi-band infrared sensors passively receive target thermal radiation information, providing all-weather sensing capabilities and enabling the identification of target thermal features under low visibility conditions. However, infrared imaging also faces challenges such as variations in atmospheric window transmittance and background thermal radiation interference, and has limitations in material identification and accurate contour extraction.

[0004] In existing technologies, multi-sensor fusion schemes typically employ decision-level fusion or feature-level fusion strategies, where information is integrated in a high-level semantic space after each sensor independently completes signal processing. While these schemes can complement the advantages of different sensors to some extent, they suffer from the following technical problems: First, the atmospheric attenuation compensation methods during the independent processing of each sensor are inconsistent, making it difficult to accurately separate common atmospheric interference from the target's independent radiation signal, resulting in residual atmospheric attenuation errors in the fused target information. Second, traditional fusion schemes lack the ability to synergistically utilize the correlation characteristics of laser and infrared modes in atmospheric transmission at the physical level, failing to fully exploit the common components of the two signals during atmospheric attenuation, thus limiting the effectiveness of atmospheric interference suppression such as defogging and haze removal. Third, in the classification and recognition of obstacles, terrain features, and passable areas, existing schemes lack an effective mechanism for jointly interpreting polarization analysis, edge gradient features, and thermal radiation information, leading to insufficient target classification accuracy and environmental understanding in complex environments.

[0005] Therefore, there is an urgent need for a UAV perception method that can achieve deep collaboration between laser and multi-band infrared at the signal acquisition level, accurately separate common atmospheric attenuation and independent target radiation at the physical model level, and jointly interpret multi-dimensional target characteristics at the perception and decision-making level, so as to improve the UAV's environmental perception capability and autonomous navigation safety in complex weather conditions. Summary of the Invention

[0006] The purpose of this application is to provide a UAV perception method and system based on laser and multi-band infrared synergy, in order to solve the problems of inaccurate separation of common atmospheric interference, insufficient synergy of multi-modal physical layers, and lack of joint interpretation capability of multi-dimensional target features in existing multi-sensor fusion schemes.

[0007] To address the aforementioned technical problems, in a first aspect, this application provides a UAV perception method based on the coordinated use of laser and multi-band infrared, comprising: The drone is equipped with lidar and multi-band infrared sensors to collect atmospheric scattering signals and infrared radiation data, respectively. Based on the atmospheric scattering signal, the atmospheric attenuation coefficient and particulate matter distribution information are obtained by inversion, and the atmospheric attenuation coefficient and particulate matter distribution information are used as atmospheric state data; based on the infrared radiation data, the background thermal distribution is analyzed, and the background thermal distribution is used as background data. A phase mask is generated based on the atmospheric state data; the phase mask is used to encode and modulate the laser echo and infrared radiation wavefront received by the UAV's optical window to obtain an intermediate image; Using the atmospheric state data and the background data as prior constraints, the intermediate image is reconstructed to obtain an initial lidar intensity map and an initial infrared thermal map. Based on the atmospheric attenuation coefficient and the particulate matter distribution information, the initial lidar intensity map and the initial infrared thermal map are jointly processed to separate the common attenuation component and the independent target radiation component, thereby obtaining the target lidar intensity map and the target infrared thermal map. The target lidar intensity map and the target infrared thermal map are input into a preset collaborative perception model. The collaborative perception model analyzes the polarization difference between the surface reflected light and the background scattered light of the target to be perceived, and jointly interprets the material, edge contour and thermal radiation information of the target to be perceived, and separates obstacles, terrain features and passable areas.

[0008] Optionally, the collaborative sensing model is configured with reflected light polarization characteristic parameters, edge gradient feature templates, and thermal radiation range data; The process involves inputting the target lidar intensity map and the target infrared thermal map into a preset collaborative sensing model. The collaborative sensing model analyzes the polarization difference between the surface reflected light and background scattered light of the target, and jointly interprets the material, edge contour, and thermal radiation information of the target to separate obstacles, terrain features, and passable areas. This includes: Extract the intensity ratio data of the target lidar intensity map in different polarization directions; The intensity ratio data is compared with the polarization characteristic parameters of the reflected light to extract pixel data belonging to the reflected light from the target surface and pixel data belonging to the background scattered light. Based on the intensity values ​​and spatial continuity of the pixel data belonging to the reflected light from the target surface, continuous edge contour data is extracted; The geometry of the edge contour data is matched with the edge gradient feature template to extract the terrain feature region; The thermal radiation values ​​of each pixel in the target infrared thermal image are compared with the thermal radiation range data to determine the material type of the object corresponding to each pixel. The regions corresponding to pixels whose thermal radiation values ​​are not within the preset ground thermal radiation range are extracted as obstacle candidate regions, and the parts that overlap with the terrain feature regions are removed from the obstacle candidate regions to obtain the obstacle regions. The area other than the obstacle candidate area and the terrain feature area, and whose surface undulation is less than a preset undulation threshold, is extracted as the passable area.

[0009] Optionally, the step of jointly processing the initial lidar intensity map and the initial infrared thermal map based on the atmospheric attenuation coefficient and the particulate matter distribution information to separate the common attenuation component and the independent target radiation component, thereby obtaining the target lidar intensity map and the target infrared thermal map, includes: The pixel values ​​of the initial lidar intensity map and the corresponding pixel values ​​of the initial infrared thermal map are obtained and combined into pixel data pairs; Based on the atmospheric attenuation coefficient and the particulate matter distribution information, the expected attenuation level at each pixel location is calculated to obtain the atmospheric influence factor. Based on the atmospheric influence factor, a common attenuation component is extracted from the pixel data pair; the common attenuation component has the same change ratio in the initial lidar intensity map and the initial infrared thermal map. The common attenuation component is removed from the initial lidar intensity map, and a first independent target radiation component reflecting the surface reflection characteristics of the target is extracted. The first independent target radiation component is then used as the target lidar intensity map. The common attenuation component is removed from the initial infrared thermal image, and a second independent target radiation component reflecting the target's own thermal radiation is extracted. The second independent target radiation component is then used as the target infrared thermal image.

[0010] Optionally, the step of calculating the expected attenuation level at each pixel location based on the atmospheric attenuation coefficient and the particulate matter distribution information to obtain the atmospheric influence factor includes: Based on the particulate matter distribution information, the relative concentration of particulate matter at the corresponding pixel location is extracted; Based on the atmospheric attenuation coefficient, the attenuation ratio of the pixel position is determined; the attenuation ratio is the ratio of the atmospheric attenuation coefficient at the laser working wavelength to the atmospheric attenuation coefficient at the infrared working wavelength. The atmospheric influence factor is obtained by multiplying the relative concentration of particulate matter with the attenuation ratio; the atmospheric influence factor is used to characterize the attenuation intensity of the atmosphere on laser echoes and infrared radiation.

[0011] Optionally, the step of reconstructing the intermediate image using the atmospheric state data and the background data as prior constraints to obtain an initial lidar intensity map and an initial infrared thermal map includes: Based on the atmospheric attenuation coefficient, the attenuation ratio corresponding to each spatial location is calculated to obtain the first constraint condition; the first constraint condition is used to limit the signal strength in the reconstruction result to not be greater than a preset physical upper limit. Based on the background thermal distribution, the background temperature data corresponding to the spatial location is extracted to obtain the second constraint condition; the second constraint condition is used to limit the infrared thermal radiation part in the reconstruction result to be no less than the background infrared radiation intensity data converted from the background temperature data. Based on the first constraint and the second constraint, calculate the estimated attenuation data and the estimated background data; The estimated attenuation data and the estimated background data are removed from the pixel data of the intermediate image to obtain the remaining signal data; The remaining signal data is used as the initial reconstruction data, and the initial reconstruction data is iteratively updated to generate a simulated intermediate image; When the difference between the simulated intermediate image and the intermediate image is less than a preset difference threshold, the initial reconstructed data at this time is used as the initial lidar intensity map and the initial infrared thermal map.

[0012] Optionally, the step of retrieving the atmospheric attenuation coefficient and particulate matter distribution information based on the atmospheric scattering signal, and using the atmospheric attenuation coefficient and particulate matter distribution information as atmospheric state data, includes: Extract the scattered light intensity value of the atmospheric scattering signal at each distance unit, and compare the scattered light intensity value with the preset standard scattering intensity to obtain scattering intensity ratio data; Calculate the rate of change of the scattering intensity ratio data with distance, and use the rate of change data as the atmospheric attenuation coefficient for the corresponding distance interval; The intensity distribution data of the atmospheric scattering signal at different scattering angles are extracted, and the relative concentration data of suspended particulate matter at each spatial location is determined based on the degree of difference of the intensity distribution data at different angles. The relative concentration data at each spatial location are combined according to spatial coordinates to obtain particulate matter distribution information; The atmospheric attenuation coefficient is combined with the particulate matter distribution information to form the atmospheric state data.

[0013] Optionally, the step of encoding and modulating the laser echo and infrared radiation wavefront received by the UAV's optical window using the phase mask to obtain an intermediate image includes: A phase mask is installed in front of the UAV's optical window. The phase mask is a transparent flat plate with a thickness undulation pattern, which is used to generate different amounts of phase delay. The laser echo and the infrared radiation wavefront are controlled to pass through the phase mask so that the wavefront phase changes according to the phase delay, thereby obtaining a modulated wavefront; the modulated wavefront has a phase difference. The modulated wavefront is transmitted to the detector array to obtain an intensity distribution map formed by interference or diffraction of the modulated wavefront; The intensity distribution map is used as the intermediate image.

[0014] Secondly, this application provides a drone perception system based on laser and multi-band infrared synergy, comprising: The data acquisition module is used to collect atmospheric scattering signals and infrared radiation data using the lidar and multi-band infrared sensors carried by the UAV, respectively. The inversion module is used to invert the atmospheric attenuation coefficient and particulate matter distribution information based on the atmospheric scattering signal, and use the atmospheric attenuation coefficient and particulate matter distribution information as atmospheric state data; and to analyze the background heat distribution based on the infrared radiation data, and use the background heat distribution as background data. The modulation module is used to generate a phase mask based on the atmospheric state data; and to use the phase mask to encode and modulate the laser echo and infrared radiation wavefront received by the UAV's optical window to obtain an intermediate image. The reconstruction module is used to reconstruct the intermediate image using the atmospheric state data and the background data as prior constraints, so as to obtain an initial lidar intensity map and an initial infrared thermal map. The separation module is used to perform joint processing on the initial lidar intensity map and the initial infrared thermal map based on the atmospheric attenuation coefficient and the particulate matter distribution information, and separate the common attenuation component and the independent target radiation component to obtain the target lidar intensity map and the target infrared thermal map. The perception module is used to input the target lidar intensity map and the target infrared thermal map into a preset collaborative perception model. The collaborative perception model analyzes the polarization difference between the surface reflected light and the background scattered light of the target to be perceived, and jointly interprets the material, edge contour and thermal radiation information of the target to be perceived, and separates obstacles, terrain features and passable areas.

[0015] Thirdly, this application provides an electronic device, comprising: Memory, used to store computer programs; A processor, used to execute the computer program to implement the steps of the UAV perception method based on laser and multi-band infrared coordination as described in the first aspect above.

[0016] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of the UAV perception method based on laser and multi-band infrared coordination as described in the first aspect above.

[0017] This application has the following beneficial effects: First, this application achieves joint encoding and modulation of laser echo and infrared radiation wavefront by generating a phase mask using atmospheric state data during the signal acquisition stage. This enables the two modal signals to be incorporated into a unified processing framework at the physical level, which is different from the traditional approach of processing each sensor independently and then performing high-level fusion. This lays the foundation for collaborative sensing from the signal source and significantly improves the consistency and accuracy of collaborative processing.

[0018] Second, this application calculates atmospheric influence factors by jointly processing the initial lidar intensity map and the initial infrared thermal map, using atmospheric attenuation coefficient and particulate matter distribution information, and accurately separates the common attenuation components caused by atmospheric attenuation and the independent radiation components of the target itself in the two modes. This effectively solves the problem of inaccurate separation of common atmospheric interference in the prior art and greatly improves the accuracy of target information extraction under complex meteorological conditions.

[0019] Third, this application organically integrates polarization analysis, edge gradient feature matching, and thermal radiation information interpretation through a collaborative perception model, establishing a multi-dimensional joint interpretation mechanism. This enables precise separation of obstacles, terrain features, and passable areas, significantly improving the UAV's environmental understanding and autonomous navigation safety in complex environments. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart illustrating a UAV perception method based on laser and multi-band infrared collaboration, provided for an embodiment of this application; Figure 2 A schematic diagram illustrating the process of atmospheric state data inversion and background data analysis provided in the embodiments of this application; Figure 3 A schematic diagram of the phase mask coding modulation process provided in the embodiments of this application; Figure 4 A schematic diagram of the process for reconstructing intermediate images with prior constraints provided in the embodiments of this application; Figure 5 A schematic diagram illustrating the process of separating common attenuation components from independent target radiation components, provided for embodiments of this application; Figure 6 A schematic diagram illustrating the process of joint interpretation of multi-dimensional features using a collaborative perception model provided in this application embodiment; Figure 7 This is a schematic diagram of the structure of a drone perception system based on the collaboration of laser and multi-band infrared, provided for an embodiment of this application. Detailed Implementation

[0022] This application provides a UAV perception method and system based on laser and multi-band infrared synergy, solving the technical problems of inaccurate separation of common atmospheric interference, insufficient multi-modal physical layer synergy, and lack of joint interpretation capability of multi-dimensional target features in existing multi-sensor fusion schemes. This application is used for autonomous environmental perception and safe navigation of UAVs in complex weather conditions.

[0023] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] The core of this application is to provide a UAV perception method based on the collaboration of laser and multi-band infrared, and a flowchart of one specific implementation is shown below. Figure 1 As shown, the method includes: S101: The UAV uses a lidar and a multi-band infrared sensor to collect atmospheric scattering signals and infrared radiation data, respectively.

[0025] LiDAR (Light Detection and Ranging) is an active optical remote sensing range sensor that acquires distance and intensity information by emitting laser pulses and receiving the reflected echoes from targets. During laser pulse transmission, particles such as aerosols, water droplets, and dust in the atmosphere scatter the laser light, forming atmospheric scattering signals. This application utilizes these atmospheric scattering signals to invert atmospheric state parameters, rather than simply filtering them out as noise.

[0026] Multi-band infrared sensors are passive thermal radiation detection devices that operate in multiple bands, including near-infrared, mid-infrared, and far-infrared. They acquire infrared radiation data by passively receiving the thermal radiation energy of target objects and the background in a scene. Different bands have different response characteristics to atmospheric windows, and multi-band configurations can provide richer information on atmospheric transmittance and target thermal radiation characteristics.

[0027] Atmospheric scattering signals refer to the optical signals formed after a laser pulse is scattered by particles such as aerosols, water droplets, and dust during its transmission through the atmosphere. These signals contain information about the physical properties of the atmospheric medium and serve as the raw data source for subsequent inversion of atmospheric attenuation coefficients and particulate matter distribution information.

[0028] It is important to note that atmospheric scattering signals and laser echoes are two different types of signals: laser echoes refer to the signal reflected back to the sensor by the target surface after the laser pulse reaches the target object, carrying information about the target's distance and reflectivity; while atmospheric scattering signals refer to the signal returned to the sensor after the laser pulse is scattered by atmospheric particles along its transmission path before reaching the target, carrying information about the physical state of the atmospheric medium itself. The two signals can be distinguished in the time domain by a distance threshold. Target echoes appear within a time window corresponding to the target distance, while atmospheric scattering signals are distributed throughout the entire time interval from the emission time to the target echo time.

[0029] Infrared radiation data refers to the scene's thermal radiation energy data received by multi-band infrared sensors in different infrared bands. It includes temperature information and radiation characteristics of the target object and the environmental background, serving as the raw data source for subsequent analysis of background thermal distribution. The acquisition of infrared radiation data covers multiple bands, and the differences between data in each band can reflect the variation characteristics of atmospheric window transmittance. This characteristic is physically related to atmospheric state data and can serve as an auxiliary reference for atmospheric compensation in subsequent steps.

[0030] It should be noted that in practical applications, the operating wavelength of lidar can be selected from commonly used bands such as 1064nm or 1550nm. The 1550nm band has advantages such as good eye safety and moderate atmospheric penetration, making it suitable for close-range perception scenarios of low-altitude UAVs. Multi-band infrared sensors cover the mid-infrared band (3-5 micrometers) and the far-infrared band (8-14 micrometers). The mid-infrared band is more sensitive to high-temperature targets, while the far-infrared band has stronger detection capabilities for normal-temperature targets. The two complement each other to cover a wider target temperature range. The UAV platform should be equipped with an inertial navigation system and a global positioning system to provide accurate pose information and ensure the spatiotemporal registration accuracy of the data collected by the two sensors. The sampling frequency of both sensors should not be lower than 10Hz to meet the timeliness requirements of real-time environmental perception during UAV flight.

[0031] For example, when a drone performs a power line inspection mission, the lidar emits a 1550nm laser pulse at a pulse repetition frequency of 10Hz. A single pulse generates approximately 3300 distance sampling units within a 500-meter detection range, with each sampling unit spaced approximately 0.15 meters apart. The intensity of the scattered echo received at each sampling unit constitutes the atmospheric scattering signal sequence corresponding to that pulse. Simultaneously, a dual-band infrared sensor acquires infrared radiation image data at a resolution of 320×256 pixels in the 3-5 micrometer and 8-14 micrometer bands, respectively.

[0032] S102. Based on the atmospheric scattering signal, the atmospheric attenuation coefficient and particulate matter distribution information are obtained by inversion, and the atmospheric attenuation coefficient and particulate matter distribution information are used as atmospheric state data; based on the infrared radiation data, the background heat distribution is analyzed, and the background heat distribution is used as background data.

[0033] It should be noted that, based on the atmospheric scattering signal and infrared radiation data output in step 101 above, this step extracts atmospheric physical state parameters and environmental background thermal radiation information from the two signals respectively, providing data support for subsequent phase mask generation, intermediate image reconstruction and separation of common attenuation components.

[0034] Furthermore, such as Figure 2 As shown, S102 specifically includes: S21. Extract the scattered light intensity value of the atmospheric scattering signal at each distance unit, and compare the scattered light intensity value with the preset standard scattering intensity to obtain scattering intensity ratio data.

[0035] The scattered light intensity value refers to the intensity of the atmospheric scattered echo signal received by the lidar at each distance sampling unit, reflecting the scattering effect of the atmospheric medium on the laser pulse at that distance. The preset standard scattering intensity refers to the reference scattering intensity value obtained under known clean atmospheric conditions, serving as a benchmark for measuring the current atmospheric scattering level. This standard value can be obtained by averaging multiple measurements of a standard target at a known distance under clear weather conditions with visibility greater than 30 kilometers. The scattering intensity ratio data refers to the ratio of the currently measured scattering intensity to the standard scattering intensity. A value greater than 1 indicates that atmospheric scattering at that distance is stronger than in clean air; a larger value indicates a higher degree of atmospheric turbidity.

[0036] S22. Calculate the rate of change of the scattering intensity ratio data with distance, and use the rate of change data as the atmospheric attenuation coefficient for the corresponding distance interval.

[0037] The atmospheric attenuation coefficient refers to the proportionality of laser signal attenuation due to atmospheric scattering and absorption per unit transmission distance. A larger value indicates stronger atmospheric absorption and scattering of the laser signal within that distance. The rate of change is calculated by performing a point-by-point difference operation on the scattering intensity ratio data along the distance direction. Specifically, the difference between the ratios of two adjacent distance units is divided by the distance interval between the two units, thus obtaining the rate of change for each distance interval. The physical significance of this difference operation is that the rate of decrease in the scattering intensity ratio with distance directly reflects the local attenuation characteristics of the laser signal by the atmosphere.

[0038] In actual calculations, to reduce noise interference, the difference results can be averaged using a sliding window, with the window width set to 5 to 10 distance units.

[0039] S23. Extract the intensity distribution data of the atmospheric scattering signal at different scattering angles, and determine the relative concentration data of suspended particulate matter at each spatial location based on the difference in intensity distribution data at different angles.

[0040] The intensity distribution data refers to the intensity distribution of the laser scattering signal at different scattering angles. Particles of different sizes and concentrations will produce characteristic scattering patterns at different scattering angles. This step is based on Mie scattering theory. When the laser wavelength and particle size are on the same order of magnitude, the scattering of laser light by particles exhibits a significant angle dependence, with forward scattering intensity being higher than backscattering intensity, and the larger the particle size, the higher the ratio of forward to backscattering. By comparing the ratio of forward to backscattering intensity, the average particle size range can be inferred; and the relative concentration of particles can be calculated from the total intensity level of the scattering signals at each angle.

[0041] The specific calculation method for the difference is as follows: the measured angular scattering intensity distribution is normalized and correlated with the preset standard scattering distribution patterns for different concentration levels. The concentration level corresponding to the standard pattern with the largest correlation coefficient is selected as the relative concentration of particulate matter at that spatial location. The preset standard scattering distribution patterns for different concentration levels are obtained in advance based on Mie scattering theory. The specific calculation parameters are as follows: assuming that the particulate matter size follows a log-normal distribution, the geometric mean particle size is set to 0.5 μm to 5 μm, the geometric standard deviation is set to 1.5 to 2.0, and the refractive index of the particulate matter in typical urban aerosol components is taken as 1.53 + 0.006i. At a laser working wavelength of 1550 nm, the normalized scattering intensity distribution is calculated every 1 degree within the range of 0 to 180 degrees using Mie scattering theory. The concentration levels are divided into 10 levels, with relative concentrations of 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.5, 2.0, 3.0, and 5.0, respectively. Each level corresponds to a set of standard scattering distribution patterns. Normalized correlation calculations use the Pearson correlation coefficient to determine the degree of linear correlation between the observed distribution and each standard model. Relative concentration data refers to the ratio of suspended particulate matter at each spatial location to the reference concentration, used to characterize the spatial density of particulate matter distribution in the atmosphere.

[0042] S24. Combine the relative concentration data of each spatial location according to spatial coordinates to obtain particulate matter distribution information.

[0043] The particulate matter distribution information refers to a three-dimensional particulate matter concentration distribution dataset indexed by spatial coordinates, reflecting the complete spatial distribution pattern of atmospheric particulate matter concentration within the UAV's sensing range. This dataset uses the UAV's current location as the origin and constructs a spatial grid along the heading, lateral, and vertical directions. Each grid node stores the relative particulate matter concentration value at the corresponding location.

[0044] S25. Combine the atmospheric attenuation coefficient with the particulate matter distribution information to form the atmospheric state data.

[0045] Atmospheric state data refers to a set of parameters that comprehensively characterize the current atmospheric optical transmission properties, including two components: atmospheric attenuation coefficient and particulate matter distribution information. The atmospheric attenuation coefficient provides the overall attenuation characteristics of the signal along the transmission path, while the particulate matter distribution information provides details of the spatial distribution of the atmospheric medium. Together, they constitute the core input data for subsequent steps in generating phase masks, calculating constraints, and calculating atmospheric influence factors.

[0046] S26. Based on the infrared radiation data, analyze the background heat distribution and use the background heat distribution as background data.

[0047] Background thermal distribution refers to the spatial distribution characteristics of steady-state thermal radiation of environmental elements such as the ground, buildings, and vegetation as reflected in infrared radiation data.

[0048] The method for analyzing background thermal distribution is as follows: Two-dimensional low-pass filtering is applied to each band of the infrared radiation data to remove high-frequency target radiation components and retain the slowly changing environmental background thermal radiation substrate. Subsequently, a weighted average is calculated for the multi-band low-pass filtering results. The weighting coefficients are determined based on the atmospheric window transmittance of each band, with higher transmittance bands assigned greater weight, thus obtaining a more accurate background thermal distribution. A Gaussian filter can be used as the low-pass filter, with its cutoff frequency set according to the spatial scale of the background region in the perceived scene. For example, when the background region is approximately 50 meters by 50 meters, the corresponding Gaussian kernel standard deviation can be set to 15 to 20 pixels. The background data serves as the basis for setting the lower limit constraint on thermal radiation in subsequent intermediate image reconstruction, ensuring that the infrared components in the reconstructed results at least include the contribution of environmental background thermal radiation.

[0049] Specifically, in one embodiment of this application, after processing the atmospheric scattering signals collected by the lidar along the flight path through the aforementioned sub-steps, the atmospheric attenuation coefficient at each point along the flight path is approximately 0.8 to 1.5 km⁻¹. The attenuation coefficient is larger in the low-altitude region due to higher particulate matter concentration, and smaller in the high-altitude region. Particulate matter distribution information shows that the relative concentration of particulate matter near the ground is approximately 1.3 to 1.8, while the relative concentration drops to 0.6 to 0.9 in the high-altitude region above 200 meters, exhibiting typical boundary layer distribution characteristics. Background thermal distribution obtained from infrared sensor analysis shows that the ground temperature is approximately 15 to 20 degrees Celsius, the temperature in the concrete pavement area is higher at approximately 22 degrees Celsius, and the temperature in the vegetation-covered area is lower at approximately 14 degrees Celsius, with different material areas exhibiting differentiated thermal radiation characteristics.

[0050] Through the above steps, this application obtains quantified atmospheric attenuation coefficient, particulate matter distribution information, and background thermal distribution, providing environmental prior data for subsequent coding modulation and image reconstruction.

[0051] S103. Generate a phase mask based on the atmospheric state data; use the phase mask to encode and modulate the laser echo and infrared radiation wavefront received by the UAV's optical window to obtain an intermediate image.

[0052] It should be noted that, based on the atmospheric state data output in step 102, this step generates a phase mask that is adapted to the current atmospheric conditions, embeds atmospheric compensation information into the imaging process itself, and performs spatial coding modulation on the incident wavefront, so that the subsequent reconstruction steps can use known coding rules to more accurately separate useful signals and interference components.

[0053] Furthermore, such as Figure 3 As shown, S103 specifically includes: S31. Install the phase mask in front of the UAV's optical window.

[0054] In this context, a phase mask refers to an optically transparent element with a specific thickness undulation pattern, installed in front of the incident light front at the optical window of a drone. The thickness undulation pattern refers to the thickness difference distribution at different locations on the phase mask plate. This difference causes the light wavefront passing through different locations to experience different optical path lengths, thus producing different phase retardations. The phase retardation refers to the phase change caused by the different thickness of the medium after the light wavefront passes through the phase mask; its magnitude is proportional to the mask thickness and the refractive index of the material.

[0055] It should be noted that the thickness undulation pattern of the phase mask is calculated and generated based on the atmospheric attenuation coefficient and particulate matter distribution information in the atmospheric state data obtained in step 102. The specific generation and calculation method is as follows: The first step is to calculate the cumulative attenuation. For a sampling point with coordinates (x, y) on the optical window, the atmospheric attenuation coefficient is linearly integrated along the corresponding spatial transmission path to obtain the cumulative attenuation A(x, y) in that direction. The formula for calculation is as follows: ; in Let be the atmospheric attenuation coefficient at a distance s along the transmission path, and L be the maximum detection range in that direction. In the discrete implementation, this integral is approximated using the trapezoidal numerical integration method.

[0056] The second step is to calculate the phase delay distribution. The cumulative attenuation is transformed into a phase delay distribution φ(x,y) through a linear mapping, using the following formula: ; in The preset maximum phase delay is 2π. This represents the maximum cumulative attenuation across all sampling points within the optical window. This linear mapping ensures that the direction with the maximum cumulative attenuation receives the greatest phase delay compensation.

[0057] The third step is to calculate the mask thickness distribution. Based on the physical relationship between the phase delay and the mask thickness, the thickness value d(x,y) at each sampling point is calculated using the following formula: ; in The reference wavelength is the laser operating wavelength, for example, 1550nm, and n is the refractive index of the mask substrate material, for example, quartz glass n=1.444.

[0058] The fourth step involves using particulate matter distribution information to refine the spatial resolution of the mask. This is done when the spatial gradient of particulate matter concentration exceeds a preset gradient threshold. In areas where the concentration gradient is less than the threshold, the sampling point spacing is reduced to half of the baseline spacing to increase the detail density of thickness undulations; in areas where the concentration gradient is less than the threshold, the baseline sampling spacing is maintained to keep the thickness distribution smooth.

[0059] It should be noted that, due to the significant difference between laser wavelength and infrared wavelength, the phase delay caused by the same thickness undulation differs for the two wavelengths. Specifically, for a mask position with thickness d, the phase delay of the laser... infrared phase delay ,in and These represent the refractive indices of the mask material at the two wavelengths. This wavelength-dependent differential phase delay causes the two modal signals to correspond to different spatial frequency components in the intermediate image, thus enabling decoupling of the two modes through frequency separation during the reconstruction stage.

[0060] In practical applications, phase masks can be fabricated on optical-grade quartz glass substrates using photolithography, with a minimum feature size of 10 micrometers and a transmittance of 95%. Under dynamic atmospheric conditions, liquid crystal spatial light modulators can be used to replace fixed quartz masks to achieve real-time updates of the phase pattern at a frequency of 1 Hz.

[0061] When using a liquid crystal spatial light modulator, the driving algorithm is as follows: At each sampling period, the atmospheric state data output in step 102 is received in real time, and the phase delay distribution is recalculated according to the above four-step calculation method. ,Will The value is converted into a liquid crystal voltage drive signal, and the conversion formula is as follows: ; in The full-range driving voltage of the liquid crystal modulator is used to update the phase pattern by writing it pixel by pixel into the liquid crystal array through a digital-to-analog converter.

[0062] S32. Control the laser echo and the infrared radiation wavefront to pass through the phase mask so that the wavefront phase changes according to the phase delay amount to obtain a modulated wavefront; the modulated wavefront has a phase difference.

[0063] In this context, the modulated wavefront refers to the wavefront signal after the laser echo and infrared radiation have been encoded by a phase mask. Its different spatial locations carry known phase-coded information introduced by the phase mask. Phase difference refers to the difference in phase values ​​between different spatial locations of the modulated wavefront due to variations in mask thickness. This difference forms the physical basis for subsequent interferometric or diffraction imaging. It should be noted that because laser echo and infrared radiation have different wavelengths, the phase delay produced when passing through a mask of the same thickness is also different. This wavelength-dependent differential phase modulation allows the intermediate image to simultaneously encode information from both modes, providing a physical basis for decomposing the intermediate image into a laser intensity map and an infrared thermogram during the subsequent reconstruction stage.

[0064] S33. The modulation wavefront is transmitted to the detector array to obtain an intensity distribution map formed by interference or diffraction of the modulation wavefront.

[0065] S34. Use the intensity distribution map as the intermediate image.

[0066] The detector array refers to the area-array photodetector in the UAV imaging system, used to receive the modulated wavefront and convert the optical signal into an electrical signal. The intensity distribution map refers to the two-dimensional light intensity distribution image formed by interference or diffraction on the detector surface after the modulated wavefront reaches the detector array, due to phase differences between different parts of the wavefront. It should be noted that interference and diffraction coexist and work together during the transmission of the modulated wavefront to the detector array: the interference effect is caused by the phase difference between light waves from different spatial locations in the modulated wavefront, forming large-scale bright and dark fringes; the diffraction effect is caused by the finite aperture edge and fine thickness undulation structure of the phase mask, forming smaller-scale diffraction spots and ring patterns. The superposition of these two effects forms an intensity distribution map containing multi-scale spatial frequency components, where low spatial frequency components are mainly contributed by the interference effect, and high spatial frequency components are mainly contributed by the diffraction effect. The intermediate image is not a direct imaging result in the traditional sense, but rather an intensity distribution after phase encoding, containing mixed encoded data of target scene information and atmospheric state information, which needs to be interpreted through subsequent calculation and reconstruction steps.

[0067] The core advantage of this coding method lies in embedding atmospheric compensation information into the imaging process itself, which enables the known coding rules to be used to more accurately separate useful signals from interference components during reconstruction, thereby improving the signal recovery quality under atmospheric interference conditions.

[0068] For example, in one embodiment of this application, the detector array has a resolution of 640 x 512 pixels and a pixel size of 15 micrometers. The intermediate image presents a characteristic pattern of superimposed interference fringes and diffraction spots with alternating brightness and darkness. Different spatial frequency components in the pattern correspond to the mixed information of the target and the atmosphere at different distances and directions.

[0069] This application encodes atmospheric state data into an optical wavefront through the above steps, forming an intermediate image containing atmospheric interference information, thus providing a data basis for subsequent signal-interference separation.

[0070] S104. Using the atmospheric state data and the background data as prior constraints, the intermediate image is reconstructed to obtain an initial lidar intensity map and an initial infrared thermal map.

[0071] It should be noted that, based on the atmospheric state data and background data output in step 102, and the intermediate image output in step 103, this step uses physical prior constraints to calculate, decode, and reconstruct the encoded intermediate image, restoring the mixed encoded data into initial image data of two independent modes: laser intensity map and infrared thermogram.

[0072] Furthermore, such as Figure 4 As shown, S104 specifically includes: S41. Based on the atmospheric attenuation coefficient, calculate the attenuation ratio corresponding to each spatial location to obtain the first constraint condition.

[0073] The attenuation ratio refers to the proportion of signal attenuation due to atmospheric transmission at each spatial location, calculated based on the atmospheric attenuation coefficient. It is calculated exponentially by multiplying the atmospheric attenuation coefficient by the signal transmission distance according to Beer-Lambert's law. The first constraint condition refers to the upper bound condition of the reconstructed signal strength derived from the attenuation ratio. That is, the reconstructed signal strength at any spatial location does not exceed the theoretical maximum strength value at that location under zero attenuation conditions multiplied by the reciprocal of the attenuation ratio. This ensures that the reconstructed signal does not exceed the physically achievable maximum strength value and avoids abnormal bright spots exceeding physical laws during the reconstruction process. The preset physical upper limit refers to the maximum theoretical signal strength value that can be reached under given atmospheric conditions and the target's maximum reflectivity.

[0074] S42. Based on the background heat distribution, extract the background temperature data corresponding to the spatial location to obtain the second constraint condition.

[0075] The background temperature data refers to the ambient background temperature values ​​extracted from the background thermal distribution for each spatial location. The second constraint is the lower bound condition for the infrared thermal radiation component in the reconstruction result. Its physical meaning is that the infrared thermal radiation of any target in the scene must be superimposed on the ambient background thermal radiation; therefore, the thermal radiation value at any location in the reconstructed infrared image should not be lower than the ambient background thermal radiation level at that location. This constraint effectively prevents non-physical variations in the infrared image caused by iterative fluctuations during the reconstruction process, which could result in local temperatures below the ambient temperature.

[0076] It should be further explained that, in terms of physical dimensions, ambient background temperature and infrared thermal radiation belong to temperature (unit: Kelvin K) and radiation emission (unit: W·m), respectively. - ²), the two cannot be directly compared in magnitude. Therefore, before applying the second constraint, the background temperature data needs to be converted into the corresponding background infrared radiation intensity data. The specific conversion method is as follows: for the environmental background temperature value at location (x,y) in the background heat distribution... According to the Stefan-Boltzmann law, the temperature value is converted into the radiative exitance of that location in the infrared operating band. The conversion formula is: Where σ is the Stefan-Boltzmann constant, and its value is... ε is the emissivity of the target surface in the infrared working band, which can be taken as 0.90 to 0.98 for general ground surfaces. The unit is Kelvin (K). The unit is W·m - ², the result This refers to the background infrared radiation intensity data. Accordingly, the specific form of the second constraint is: the infrared thermal radiation value at position (x, y) in the reconstruction result. satisfy This ensures that both ends of the second constraint are in the dimension of radiative emission, avoiding the dimensional inconsistency problem that arises from directly comparing temperature and radiation. When the infrared sensor outputs the integrated radiance of a specified band, the above Stefan-Boltzmann full-band conversion can be replaced with the definite integral form of the Planck blackbody radiation function with respect to the wavelength for the corresponding band. The emissivity ε is then taken as the equivalent emissivity of that band, thus ensuring that the background infrared radiation intensity data and the infrared thermogram use the same band aperture and unit.

[0077] S43. Calculate the estimated attenuation data and estimated background data based on the first constraint and the second constraint.

[0078] The estimated attenuation data refers to the initial estimate of the atmospheric attenuation contribution at each pixel location, calculated based on the first constraint. This is calculated by integrating the atmospheric attenuation coefficient over the signal transmission path in each direction to obtain an estimate of the energy loss due to atmospheric absorption and scattering at each pixel location. The estimated background data refers to the initial estimate of the environmental background thermal radiation contribution at each pixel location, calculated based on the second constraint. This is directly taken from the temperature value at the corresponding location in the background thermal distribution and converted into radiation intensity. Together, these two data constitute the initial estimate of the known interference components in the intermediate image.

[0079] S44. Remove the estimated attenuation data and the estimated background data from the pixel data of the intermediate image to obtain the remaining signal data.

[0080] The remaining signal data refers to the signal data retained after subtracting known interference components from the intermediate image. It mainly contains useful information about the target scene and estimation error components, and serves as the initial input for iterative reconstruction. The removal operation is performed at the pixel level, that is, subtracting the estimated attenuation value and estimated background value at the corresponding position from each pixel value in the intermediate image.

[0081] S45. Use the remaining signal data as initial reconstruction data, and iteratively update the initial reconstruction data to generate a simulated intermediate image.

[0082] The initial reconstructed data refers to the first set of data in the reconstructed data sequence, starting from the remaining signal data. It should be noted that the initial reconstructed data contains two components: a laser intensity component and an infrared thermal radiation component. The separation of these two components is based on the wavelength-dependent phase difference introduced by the coding modulation in step 103. Because the laser wavelength and infrared wavelength are different, a mask of the same thickness produces different phase delays for both. Therefore, the signals of the two modes in the intermediate image correspond to different spatial frequency components and interference fringe spacings. This frequency difference can be used to decompose the initial reconstructed data into two modal components.

[0083] Iterative updates refer to the process in each iteration of generating a simulated intermediate image using the forward propagation model based on the current laser intensity component and infrared thermal radiation component, comparing it with the actual intermediate image, and adjusting the two components in reverse according to the error. The forward propagation model simulates the complete physical process of a signal traveling from the target through the atmosphere, passing through phase mask coding and modulation, and finally reaching the detector array to form an intensity distribution map.

[0084] In each iteration, the current component is first checked to see if it satisfies the first and second constraints. If not, it is forcibly projected into the constraint domain, and then forward propagation calculation is performed. The simulated intermediate image refers to the theoretical intermediate image that should be generated after phase mask coding modulation, based on the current reconstruction result.

[0085] S46. When the difference between the simulated intermediate image and the intermediate image is less than a preset difference threshold, the initial reconstructed data at this time is used as the initial lidar intensity map and the initial infrared thermal map.

[0086] Here, the difference degree refers to the quantitative deviation between the simulated intermediate image and the actual intermediate image, which can be calculated using indicators such as mean square error or relative error. The preset difference threshold is the criterion for determining iterative convergence. When the difference degree decreases below this threshold, the reconstruction result is considered sufficiently accurate, and the iteration terminates. The initial lidar intensity map is the laser intensity component of the reconstructed data at convergence, and the initial infrared thermal map is the infrared thermal radiation component of the reconstructed data at convergence. At this point, the two images have basically restored the image information of the target scene in two modes, but they still contain the influence of atmospheric attenuation, which needs to be further separated and removed in the subsequent step 105.

[0087] In practical implementation, the preset difference threshold can be set to a mean square error of less than 0.01 or a relative error of less than 1%, and the upper limit of the number of iterations can be set to 500. Iterative updates can be implemented using convex optimization algorithms such as the alternating direction multiplier method or the near-end gradient descent method. In one embodiment of this application, convergence occurs after approximately 200 iterations under hazy conditions. The clarity of the initial lidar intensity map and the initial infrared thermal map is significantly improved compared to direct imaging. The originally blurred target outline in the lidar intensity map is restored to clear spatial details, and the spatial resolution of the temperature distribution in the infrared thermal map is improved from approximately 2 meters to approximately 0.5 meters.

[0088] Through the above steps, this application uses atmospheric state data and background data as prior constraints to initially reconstruct an initial image that has eliminated some environmental interference from the coded and modulated intermediate image, providing a foundation for subsequent accurate separation.

[0089] S105. Based on the atmospheric attenuation coefficient and the particulate matter distribution information, the initial lidar intensity map and the initial infrared thermal map are jointly processed to separate the common attenuation component and the independent target radiation component, thereby obtaining the target lidar intensity map and the target infrared thermal map.

[0090] It should be noted that, based on the atmospheric attenuation coefficient and particulate matter distribution information output in step 102, and the initial lidar intensity map and initial infrared thermal map output in step 104, this step utilizes the physical correlation between the two modes in atmospheric transmission to accurately separate the common interference component caused by atmospheric attenuation and the independent radiation signal of the target itself, providing high-quality target feature data for the subsequent collaborative sensing model.

[0091] Furthermore, such as Figure 5 As shown, S105 specifically includes: S51. Obtain the pixel values ​​of the initial lidar intensity map and the corresponding pixel values ​​of the initial infrared thermal map, and combine them into pixel data pairs.

[0092] Here, a pixel data pair refers to a binary combination of laser intensity and infrared thermal radiation values ​​at the same spatial location. Each pixel data pair contains synchronous observations from two different modal sensors at that location and serves as the basic analytical unit for subsequent separation of common attenuation components. Spatial registration of the two images is ensured by the inertial navigation and positioning system in step 101, with a pixel-level alignment error not exceeding 0.5 pixels.

[0093] S52. Based on the atmospheric attenuation coefficient and the particulate matter distribution information, calculate the expected attenuation level at each pixel location to obtain the atmospheric influence factor.

[0094] It should be noted that the calculation process for atmospheric impact factors includes the following three steps: First, based on particulate matter distribution information, the relative concentration of particulate matter at the corresponding pixel location is extracted. The relative concentration of particulate matter refers to the ratio of the particulate matter concentration at a specific pixel location extracted from the particulate matter distribution information to the reference concentration, reflecting the actual density of atmospheric particulate matter at that location.

[0095] Second, based on the atmospheric attenuation coefficient, the attenuation ratio of the pixel position is determined. This attenuation ratio is the ratio of the atmospheric attenuation coefficient at the laser operating wavelength to the atmospheric attenuation coefficient at the infrared operating wavelength. It should be further clarified that this ratio is not a direct ratio of the laser wavelength to the infrared wavelength, but rather a measured value of the atmospheric attenuation coefficient at the laser operating wavelength (e.g., 1550nm). Measured values ​​of atmospheric attenuation coefficient at infrared operating wavelength (e.g., 10 micrometers) The ratio, i.e., the attenuation ratio The physical significance of the attenuation ratio lies in the fact that atmospheric particles have different scattering and absorption efficiencies for electromagnetic waves of different wavelengths. According to Mie scattering theory, when the particle size and light wavelength are on the same order of magnitude, the scattering effect on short-wavelength signals is usually stronger than that on long-wavelength signals. Therefore, the attenuation intensity experienced by a 1550nm laser and a 10-micron infrared signal under the same atmospheric particulate conditions has a fixed proportional relationship, which can be determined by the ratio of the atmospheric attenuation coefficient values ​​at the two wavelengths.

[0096] Third, the atmospheric influence factor is obtained by multiplying the relative concentration of particulate matter with the attenuation ratio. The atmospheric influence factor integrates information from two dimensions: the spatial variation of particulate matter concentration and the wavelength-dependent attenuation ratio. It is used to characterize the actual attenuation intensity of the atmosphere on laser echo and infrared radiation at each pixel location and is a key parameter for separating common attenuation components.

[0097] For example, under conditions of 1550nm laser wavelength and 10μm infrared wavelength, assuming the relative concentration of particulate matter at a certain pixel location is 1.2 and the attenuation ratio between laser and infrared is 0.85, then the atmospheric influence factor at that location is 1.2 multiplied by 0.85, which equals 1.02. At another location, the relative concentration of particulate matter is 0.7, so the atmospheric influence factor is 0.7 multiplied by 0.85, which equals 0.595, reflecting weaker atmospheric attenuation at that location.

[0098] S53. Based on the atmospheric influence factor, extract the common attenuation component from the pixel data pair.

[0099] The common attenuation component refers to the signal attenuation caused simultaneously by atmospheric scattering and absorption in both the laser intensity map and the infrared thermogram. Its core characteristic is that it exhibits the same variation ratio in both modes. The physical meaning of this variation ratio is that atmospheric particles attenuate both laser and infrared signals traveling along the same spatial path. Although the absolute attenuation amounts differ between the two wavelengths, the spatial distribution pattern of the attenuation is consistent. Therefore, by establishing the coupling relationship between the signal attenuation of the two modes using atmospheric influence factors, this common component can be decomposed from pixel data pairs.

[0100] Optionally, one specific separation method is as follows: Let the laser observation value at a certain pixel position in the pixel data pair be... Infrared observation value If the atmospheric influence factor at this location is k, then the laser target signal is The infrared target signal is The common attenuation satisfies the constraint Substituting the above relationships, we construct the linear least squares objective function. , where c= Let J be the infrared attenuation component to be determined. Taking the derivative of J with respect to c and setting the derivative to zero, we obtain the analytical solution. ,in and Let c be the current estimated value of the target signal, which can be initially set to zero. From this, the infrared component in the common attenuation component is denoted as c, and the laser component as c. .

[0101] S54. Remove the common attenuation component from the initial lidar intensity map, extract the first independent target radiation component that reflects the surface reflection characteristics of the target, and use the first independent target radiation component as the target lidar intensity map.

[0102] The first independent target radiation component refers to the signal retained after subtracting the common attenuation component from the initial lidar intensity map. This signal accurately reflects the laser reflection characteristics of the target surface and is no longer affected by atmospheric attenuation. The removal operation is performed at the pixel level, that is, subtracting the laser component from the common attenuation component corresponding to each pixel value in the initial lidar intensity map. The target lidar intensity map refers to the laser intensity image after removing common atmospheric attenuation.

[0103] S55. Remove the common attenuation component from the initial infrared thermal image, extract the second independent target radiation component that reflects the target's own thermal radiation, and use the second independent target radiation component as the target infrared thermal image.

[0104] The second independent target radiation component refers to the signal retained after subtracting the common attenuation component from the initial infrared thermal image. This signal truly reflects the target's own thermal radiation characteristics and is no longer affected by atmospheric attenuation. The removal method is the same as S54, that is, subtracting the infrared component from the common attenuation component from each pixel value of the initial infrared thermal image. The target infrared thermal image refers to the infrared thermal radiation image after removing common atmospheric attenuation.

[0105] After the above processing, the common atmospheric attenuation in the target lidar intensity map and the target infrared thermal map has been effectively removed, and the retained signal can more realistically reflect the physical characteristics of the target itself. In one embodiment of this application, after the common attenuation component is separated, the signal-to-noise ratio of the reflection characteristics of the power tower in the target lidar intensity map is improved by about 8dB, and the reflection details of the tower's metal structure surface, which were originally blurred due to atmospheric scattering, are clearly presented. The thermal radiation characteristics of the transmission line in the target infrared thermal map are improved by about 40% from the contrast in the background, and the temperature difference characteristics between the line and the surrounding air are more prominent.

[0106] Through the above steps, this application separates the common attenuation component caused by atmospheric attenuation from the independent radiation component of the target itself, and obtains a true target laser intensity map and infrared thermal map that eliminates the influence of the atmosphere, providing clean target feature data for subsequent collaborative sensing.

[0107] S106. Input the target lidar intensity map and the target infrared thermal map into a preset collaborative perception model. The collaborative perception model analyzes the polarization difference between the surface reflected light and the background scattered light of the target to be perceived, and jointly interprets the material, edge contour and thermal radiation information of the target to be perceived, and separates obstacles, terrain features and passable areas.

[0108] It should be noted that, based on the target lidar intensity map and target infrared thermal map output in step 105, this step uses a collaborative perception model to perform multi-dimensional feature joint interpretation on the dual-modal data after removing common atmospheric attenuation, ultimately achieving accurate separation of obstacles, terrain features and passable areas, providing environmental understanding support for UAV autonomous navigation.

[0109] Figure 6 The flowchart shows the sub-steps of the collaborative perception model for joint interpretation of multi-dimensional features provided in the embodiments of this application.

[0110] The collaborative sensing model refers to a multi-dimensional feature joint interpretation model that pre-configures reflected light polarization characteristic parameters, edge gradient feature templates, and thermal radiation range data. It is used for comprehensive analysis and processing of target lidar intensity maps and target infrared thermal maps. The collaborative sensing model is a deterministic rule-matching processing model based on a pre-set parameter database, requiring no training process. Its parameter database was constructed through the calibration experiments described below.

[0111] Among them, the polarization characteristic parameter of reflected light refers to a pre-calibrated database of differences in polarization characteristics between reflected light from different types of target surfaces and atmospheric background scattered light, including reference ranges of intensity ratios of various typical target materials in different polarization directions.

[0112] The database was constructed as follows: In a laboratory environment, using a 1550nm polarized laser light source, polarization reflectance measurements were performed on five types of materials—metal, rock, vegetation, water surface, and concrete—at incident angles of 30°, 40°, 50°, 60°, and 70°. For metal, three samples each of stainless steel and aluminum alloy were used; for rock, three samples each of granite and sandstone; for vegetation, three samples each of grassland and foliage; for water surface, three samples each of still water and microwave water; and for concrete, three samples each of ordinary concrete and asphalt. The measurement environment was controlled as follows: temperature 20±2°C, relative humidity 50±10%, and no external light interference. At least 50 repeated measurements were performed for each material at each angle. The mean and standard deviation of the ratio of horizontal polarized reflectance intensity to vertical polarized reflectance intensity were used to form a reference range for the polarization ratio of each material at each angle. The polarization characteristics of atmospheric background scattered light were obtained by performing the same measurements on sky scattered light without a target direction.

[0113] Among them, the edge gradient feature template refers to a set of templates containing the geometric contours of typical terrain features, including gradient distribution templates of characteristic shapes such as ridgelines, gully edges, building corners, and road boundaries. Each template is described by the joint distribution of gradient direction and gradient magnitude, supporting matching at different scales and rotation angles. The feature vector representation of each template is as follows: N=36 sampling points are taken at equal intervals on the edge contour. A logarithmic polar coordinate system is established with each sampling point as the center. The radius is normalized to three levels, corresponding to 0.125, 0.25, and 0.5 times the total contour length, respectively. The angle is evenly divided into 12 directions. The number of other sampling points falling within each polar coordinate grid is counted to form a 36-dimensional histogram feature vector.

[0114] Among them, the thermal radiation range data refers to the range of thermal radiation values ​​for different material types within common temperature ranges.

[0115] For example, at an ambient temperature of 20 degrees Celsius, the infrared radiation value of a metallic material is approximately 8 to 12. The concrete material is approximately 14 to 18. The vegetation is approximately 16 to 22 The water body is approximately 18 to 24. The preset ground thermal radiation range incorporates the radiation range of common ground cover materials such as soil, rocks, and vegetation, and is set to 12 to 25. It should be noted that the above radiation range data depends on ambient temperature. When the ambient temperature changes, the thermal radiation range data is corrected using the following temperature compensation mechanism: the radiation value of each material at any ambient temperature T is calculated according to the Stefan-Boltzmann law... Calculation, where The surface emissivity of each material, for example, that of metals. =0.15 to 0.30, concrete =0.85 to 0.95, vegetation =0.90 to 0.98, water body =0.95 to 0.98, soil =0.85 to 0.95, Stefan-Boltzmann constant T is the Kelvin temperature. The current ambient temperature T is obtained from the background heat distribution data analyzed in step 102. It is then substituted into the above formula to calculate the radiation range of each material at the current temperature, thereby achieving adaptive temperature compensation for the thermal radiation range data.

[0116] Furthermore, step 106 may include the following sub-steps: S61. Extract the intensity ratio data of the target lidar intensity map in different polarization directions.

[0117] The intensity ratio data refers to the ratio of the intensity of the target lidar intensity map in the horizontal polarization direction to that in the vertical polarization direction. Its physical basis is Fresnel's law of reflection, which states that the reflectivity of the horizontal and vertical polarization components differs under different incident angles and surface materials. Different types of targets exhibit different intensity ratio distributions in the polarization direction due to differences in surface microstructure and material properties. Metallic surfaces typically have a higher degree of polarization, with ratios deviating significantly from 1; rough surfaces have a lower degree of polarization, with ratios close to 1.

[0118] S62. Compare the intensity ratio data with the polarization characteristic parameters of the reflected light, and extract the pixel data belonging to the reflected light from the target surface and the pixel data belonging to the background scattered light.

[0119] This step utilizes the physical characteristic that target surface reflected light typically has a strong degree of polarization, while atmospheric background scattered light has a low degree of polarization. The specific comparison method is as follows: a polarization degree threshold is set; pixels with a polarization degree higher than this threshold are assigned to the target surface reflected light set, and pixels with a polarization degree lower than this threshold are assigned to the background scattered light set. The polarization degree threshold can be determined statistically based on the polarization degree boundary between the target material and atmospheric scattering from a reflected light polarization characteristic parameter database, with a typical value of 0.15 to 0.25.

[0120] S63. Extract continuous edge contour data based on the intensity value and spatial continuity of the pixel data belonging to the reflected light of the target surface.

[0121] Spatial continuity refers to the continuous spatial distribution of pixels reflecting light from the target surface. Spatial continuity constraints ensure that the extracted edges are genuine target edges, not discrete bright spots caused by noise. In practice, gradient calculations are first performed on the pixels reflecting light from the target surface, and pixels with gradient magnitudes higher than a threshold are extracted as edge candidate points. Then, connected component analysis is performed on the candidate points, retaining chains of edge candidate points whose connected component lengths exceed a preset minimum length threshold, forming continuous edge contour data. The preset minimum length threshold can be set to 10 pixels to filter out short fragments caused by noise. Edge contour data refers to the set of target edge pixels and their spatial geometric shape description data after continuity filtering.

[0122] S64. Match the geometry of the edge contour data with the edge gradient feature template to extract the terrain feature region.

[0123] Among them, the terrain feature region refers to the region with terrain feature attributes that is confirmed by matching the edge geometry with typical terrain features in the template library, including terrain elements such as ridges, gullies, and slope transitions.

[0124] The matching process uses the shape context descriptor method, taking 36 sampling points at equal intervals on the edge contour to be matched and the template contour respectively, calculating the 36-dimensional log-polar histogram feature vector of each sampling point, and then calculating the cosine similarity between the two sets of feature vectors. When the average cosine similarity exceeds the preset matching threshold of 0.8, it is determined to be the terrain feature of the corresponding type.

[0125] S65. Compare the thermal radiation values ​​of each pixel in the target infrared thermal image with the thermal radiation range data to determine the material type of the object corresponding to each pixel.

[0126] Different materials exhibit different thermal radiation characteristics at the same ambient temperature due to differences in specific heat capacity, thermal conductivity, and surface emissivity. Therefore, the comparison method is as follows: the thermal radiation value of each pixel is compared with the radiation range of each material type in the thermal radiation range data to determine the interval. When a pixel's radiation value falls within the radiation range of a certain material, the pixel is determined to be of the corresponding material type. When a pixel's radiation value falls within the overlapping intervals of multiple material ranges, the reflection intensity value of the pixel in the target lidar intensity map is used for auxiliary determination: the absolute value of the difference between the pixel's reflection intensity value and the typical reflectivity of each candidate material is calculated, and the material type with the smallest difference is selected as the final determination result. The material type refers to the material classification result of the object corresponding to the pixel determined through the above comparison, including categories such as metal, concrete, vegetation, water, and soil.

[0127] S66. Extract the regions corresponding to pixels whose thermal radiation values ​​are not within the preset ground thermal radiation range as obstacle candidate regions, and remove the parts that overlap with the terrain feature regions from the obstacle candidate regions to obtain the obstacle regions.

[0128] The preset ground thermal radiation range refers to the range of thermal radiation values ​​of the natural ground environment under normal temperature conditions. Pixels with thermal radiation values ​​exceeding this range are considered to potentially correspond to non-ground objects. For example, man-made targets such as metal equipment and vehicles typically have radiation values ​​below the lower limit of the ground range, while operating electrical equipment has radiation values ​​above the upper limit. The obstacle candidate region refers to the spatial area formed by the set of pixels with abnormal thermal radiation values. The purpose of removing overlapping parts of terrain feature regions is to avoid misclassifying areas of abrupt changes in natural terrain as man-made obstacles. The obstacle region refers to the finally confirmed obstacle spatial region.

[0129] S67. Extract the area other than the obstacle candidate area and the terrain feature area, and whose surface undulation is less than a preset undulation threshold, as the passable area.

[0130] Surface undulation refers to the local elevation difference calculated based on LiDAR point cloud data, specifically the difference between the maximum and minimum elevation values ​​within a preset window centered on each pixel. The preset undulation threshold is a flatness judgment standard set according to the UAV type and mission requirements. For low-altitude inspection missions of multi-rotor UAVs, the undulation threshold can be set to 0.5 meters; for long-range cruise missions of fixed-wing UAVs, it can be relaxed to 2 meters. The passable area refers to an area where obstacles and terrain features are eliminated, and the ground is sufficiently flat, allowing UAVs to safely pass through or land.

[0131] For example, the application of step 106 is illustrated by taking a drone performing a power line inspection task in smoggy weather. The collaborative perception model processing results show that the power tower, due to its metallic material, has a polarization degree of 0.35 in polarization analysis, far exceeding the 0.08 to 0.12 of the background scattered light, and is clearly classified into the target surface reflected light set; in thermal radiation analysis, the radiation value of the metallic tower is approximately 9. 12 below the lower limit of the ground thermal radiation range The ridgeline edges along the power line were extracted as candidate obstacle regions. The similarity between the ridgeline outlines and the ridge templates in the template library reached 0.87, and they were effectively extracted as terrain feature regions. The surface undulation of the flat farmland area below the power lines was only 0.15 meters, far below the 0.5-meter threshold, and was identified as a passable area. The entire processing achieved accurate separation of obstacles, terrain, and passable areas in the power line inspection scenario, providing reliable environmental understanding support for the safe autonomous flight of UAVs.

[0132] Through the above steps, this application utilizes a collaborative perception model to comprehensively analyze the polarization characteristics, edge contours, and thermal radiation information of the target, ultimately achieving accurate separation and identification of obstacles, terrain features, and passable areas.

[0133] This application, through steps S101-S106, realizes a complete processing flow from atmospheric interference acquisition, environmental parameter inversion, coding and modulation, prior constraint reconstruction, common attenuation separation to collaborative perception classification, effectively eliminating the impact of atmospheric scattering and attenuation on perception data, and improving the accuracy and reliability of UAVs in identifying obstacles, terrain and passable areas under complex meteorological conditions.

[0134] Figure 7 This application provides a schematic diagram of a specific implementation of a UAV perception system based on laser and multi-band infrared collaboration, referring to... Figure 7 The system may include: The data acquisition module 71 is used to acquire atmospheric scattering signals and infrared radiation data respectively through the lidar and multi-band infrared sensor carried by the UAV. The inversion module 72 is used to invert the atmospheric attenuation coefficient and particulate matter distribution information based on the atmospheric scattering signal, and use the atmospheric attenuation coefficient and particulate matter distribution information as atmospheric state data; and to analyze the background heat distribution based on the infrared radiation data, and use the background heat distribution as background data. Modulation module 73 is used to generate a phase mask based on the atmospheric state data; and to encode and modulate the laser echo and infrared radiation wavefront received by the UAV optical window using the phase mask to obtain an intermediate image. The reconstruction module 74 is used to reconstruct the intermediate image using the atmospheric state data and the background data as prior constraints, to obtain an initial lidar intensity map and an initial infrared thermal map. The separation module 75 is used to perform joint processing on the initial lidar intensity map and the initial infrared thermal map based on the atmospheric attenuation coefficient and the particulate matter distribution information, to separate the common attenuation component and the independent target radiation component, and to obtain the target lidar intensity map and the target infrared thermal map. The perception module 76 is used to input the target lidar intensity map and the target infrared thermal map into a preset collaborative perception model. The collaborative perception model analyzes the polarization difference between the surface reflected light and the background scattered light of the target to be perceived, and jointly interprets the material, edge contour and thermal radiation information of the target to be perceived, and separates obstacles, terrain features and passable areas.

[0135] The UAV perception system based on laser and multi-band infrared coordination in this application is used to implement the aforementioned UAV perception method based on laser and multi-band infrared coordination. Therefore, the specific implementation of the UAV perception system based on laser and multi-band infrared coordination can be found in the embodiment section of the UAV perception method based on laser and multi-band infrared coordination mentioned above. The specific implementation can be referred to the description of the corresponding embodiments, which will not be repeated here.

[0136] This application also provides an electronic device, including: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of any of the above-described UAV perception methods based on laser and multi-band infrared coordination.

[0137] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described UAV perception methods based on laser and multi-band infrared coordination.

[0138] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0139] The embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above embodiments of the UAV perception method based on laser and multi-band infrared coordination.

[0140] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0141] The foregoing has provided a detailed description of a UAV perception method and system based on laser and multi-band infrared coordination. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A UAV perception method based on laser and multi-band infrared synergy, characterized in that, include: The drone is equipped with lidar and multi-band infrared sensors to collect atmospheric scattering signals and infrared radiation data, respectively. Based on the atmospheric scattering signal, the atmospheric attenuation coefficient and particulate matter distribution information are obtained by inversion, and the atmospheric attenuation coefficient and particulate matter distribution information are used as atmospheric state data; based on the infrared radiation data, the background thermal distribution is analyzed, and the background thermal distribution is used as background data. A phase mask is generated based on the atmospheric state data; The phase mask is used to encode and modulate the laser echo and infrared radiation wavefront received by the UAV's optical window to obtain an intermediate image. Using the atmospheric state data and the background data as prior constraints, the intermediate image is reconstructed to obtain an initial lidar intensity map and an initial infrared thermal map. Based on the atmospheric attenuation coefficient and the particulate matter distribution information, the initial lidar intensity map and the initial infrared thermal map are jointly processed to separate the common attenuation component and the independent target radiation component, thereby obtaining the target lidar intensity map and the target infrared thermal map. The target lidar intensity map and the target infrared thermal map are input into a preset collaborative perception model. The collaborative perception model analyzes the polarization difference between the surface reflected light and the background scattered light of the target to be perceived, and jointly interprets the material, edge contour and thermal radiation information of the target to be perceived, and separates obstacles, terrain features and passable areas.

2. The method according to claim 1, characterized in that, The collaborative sensing model is configured with reflected light polarization characteristic parameters, edge gradient feature templates, and thermal radiation range data. The process involves inputting the target lidar intensity map and the target infrared thermal map into a preset collaborative sensing model. The collaborative sensing model analyzes the polarization difference between the surface reflected light and background scattered light of the target, and jointly interprets the material, edge contour, and thermal radiation information of the target to separate obstacles, terrain features, and passable areas. This includes: Extract the intensity ratio data of the target lidar intensity map in different polarization directions; The intensity ratio data is compared with the polarization characteristic parameters of the reflected light to extract pixel data belonging to the reflected light from the target surface and pixel data belonging to the background scattered light. Based on the intensity values ​​and spatial continuity of the pixel data belonging to the reflected light from the target surface, continuous edge contour data is extracted; The geometry of the edge contour data is matched with the edge gradient feature template to extract the terrain feature region; The thermal radiation values ​​of each pixel in the target infrared thermal image are compared with the thermal radiation range data to determine the material type of the object corresponding to each pixel. The regions corresponding to pixels whose thermal radiation values ​​are not within the preset ground thermal radiation range are extracted as obstacle candidate regions, and the parts that overlap with the terrain feature regions are removed from the obstacle candidate regions to obtain the obstacle regions. The area other than the obstacle candidate area and the terrain feature area, and whose surface undulation is less than a preset undulation threshold, is extracted as the passable area.

3. The method according to claim 1, characterized in that, The step of jointly processing the initial lidar intensity map and the initial infrared thermal map based on the atmospheric attenuation coefficient and the particulate matter distribution information, separating the common attenuation component and the independent target radiation component, and obtaining the target lidar intensity map and the target infrared thermal map includes: The pixel values ​​of the initial lidar intensity map and the corresponding pixel values ​​of the initial infrared thermal map are obtained and combined into pixel data pairs; Based on the atmospheric attenuation coefficient and the particulate matter distribution information, the expected attenuation level at each pixel location is calculated to obtain the atmospheric influence factor. Based on the atmospheric influence factor, a common attenuation component is extracted from the pixel data pair; the common attenuation component has the same change ratio in the initial lidar intensity map and the initial infrared thermal map. The common attenuation component is removed from the initial lidar intensity map, and a first independent target radiation component reflecting the surface reflection characteristics of the target is extracted. The first independent target radiation component is then used as the target lidar intensity map. The common attenuation component is removed from the initial infrared thermal image, and a second independent target radiation component reflecting the target's own thermal radiation is extracted. The second independent target radiation component is then used as the target infrared thermal image.

4. The method according to claim 3, characterized in that, The step of calculating the expected attenuation level at each pixel location based on the atmospheric attenuation coefficient and the particulate matter distribution information to obtain the atmospheric influence factor includes: Based on the particulate matter distribution information, the relative concentration of particulate matter at the corresponding pixel location is extracted; Based on the atmospheric attenuation coefficient, the attenuation ratio of the pixel position is determined; the attenuation ratio is the ratio of the atmospheric attenuation coefficient at the laser working wavelength to the atmospheric attenuation coefficient at the infrared working wavelength. The atmospheric influence factor is obtained by multiplying the relative concentration of particulate matter with the attenuation ratio; the atmospheric influence factor is used to characterize the attenuation intensity of the atmosphere on laser echoes and infrared radiation.

5. The method according to claim 1, characterized in that, The process of reconstructing the intermediate image using the atmospheric state data and the background data as prior constraints to obtain an initial lidar intensity map and an initial infrared thermal map includes: Based on the atmospheric attenuation coefficient, the attenuation ratio corresponding to each spatial location is calculated to obtain the first constraint condition; the first constraint condition is used to limit the signal strength in the reconstruction result to not be greater than a preset physical upper limit. Based on the background thermal distribution, the background temperature data corresponding to the spatial location is extracted to obtain the second constraint condition; the second constraint condition is used to limit the infrared thermal radiation part in the reconstruction result to be no less than the background infrared radiation intensity data converted from the background temperature data. Based on the first constraint and the second constraint, calculate the estimated attenuation data and the estimated background data; The estimated attenuation data and the estimated background data are removed from the pixel data of the intermediate image to obtain the remaining signal data; The remaining signal data is used as the initial reconstruction data, and the initial reconstruction data is iteratively updated to generate a simulated intermediate image; When the difference between the simulated intermediate image and the intermediate image is less than a preset difference threshold, the initial reconstructed data at this time is used as the initial lidar intensity map and the initial infrared thermal map.

6. The method according to claim 1, characterized in that, The step of retrieving atmospheric attenuation coefficient and particulate matter distribution information based on the atmospheric scattering signal, and using the atmospheric attenuation coefficient and particulate matter distribution information as atmospheric state data, includes: Extract the scattered light intensity value of the atmospheric scattering signal at each distance unit, and compare the scattered light intensity value with the preset standard scattering intensity to obtain scattering intensity ratio data; Calculate the rate of change of the scattering intensity ratio data with distance, and use the rate of change data as the atmospheric attenuation coefficient for the corresponding distance interval; The intensity distribution data of the atmospheric scattering signal at different scattering angles are extracted, and the relative concentration data of suspended particulate matter at each spatial location is determined based on the degree of difference of the intensity distribution data at different angles. The relative concentration data at each spatial location are combined according to spatial coordinates to obtain particulate matter distribution information; The atmospheric attenuation coefficient is combined with the particulate matter distribution information to form the atmospheric state data.

7. The method according to claim 1, characterized in that, The process of encoding and modulating the laser echo and infrared radiation wavefront received by the UAV's optical window using the phase mask to obtain an intermediate image includes: A phase mask is installed in front of the UAV's optical window. The phase mask is a transparent flat plate with a thickness undulation pattern, which is used to generate different amounts of phase delay. The laser echo and the infrared radiation wavefront are controlled to pass through the phase mask so that the wavefront phase changes according to the phase delay, thereby obtaining a modulated wavefront; the modulated wavefront has a phase difference. The modulated wavefront is transmitted to the detector array to obtain an intensity distribution map formed by interference or diffraction of the modulated wavefront; The intensity distribution map is used as the intermediate image.

8. A drone perception system based on laser and multi-band infrared synergy, characterized in that, include: The data acquisition module is used to collect atmospheric scattering signals and infrared radiation data using the lidar and multi-band infrared sensors carried by the UAV, respectively. The inversion module is used to invert the atmospheric attenuation coefficient and particulate matter distribution information based on the atmospheric scattering signal, and use the atmospheric attenuation coefficient and particulate matter distribution information as atmospheric state data; and to analyze the background heat distribution based on the infrared radiation data, and use the background heat distribution as background data. A modulation module is used to generate a phase mask based on the atmospheric state data; The phase mask is used to encode and modulate the laser echo and infrared radiation wavefront received by the UAV's optical window to obtain an intermediate image. The reconstruction module is used to reconstruct the intermediate image using the atmospheric state data and the background data as prior constraints, so as to obtain an initial lidar intensity map and an initial infrared thermal map. The separation module is used to perform joint processing on the initial lidar intensity map and the initial infrared thermal map based on the atmospheric attenuation coefficient and the particulate matter distribution information, and separate the common attenuation component and the independent target radiation component to obtain the target lidar intensity map and the target infrared thermal map. The perception module is used to input the target lidar intensity map and the target infrared thermal map into a preset collaborative perception model. The collaborative perception model analyzes the polarization difference between the surface reflected light and the background scattered light of the target to be perceived, and jointly interprets the material, edge contour and thermal radiation information of the target to be perceived, and separates obstacles, terrain features and passable areas.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the UAV perception method based on laser and multi-band infrared coordination as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the implementation of the UAV perception method based on laser and multi-band infrared coordination as described in any one of claims 1 to 7.

Images