Target reflectivity-based tiny unmanned aerial vehicle detection method and system

Abstract

The invention discloses a tiny unmanned aerial vehicle detection method and system based on target reflectivity, solves the problem that a low-reflectivity target is difficult to detect in a complex environment in the prior art, and realizes recognition of a tiny unmanned aerial vehicle by combining laser active detection and target reflectivity characteristic analysis. The method comprises the following steps: transmitting pulse laser to a target area, and synchronously acquiring a target signal through double detectors with a common-path design to obtain a target geometric feature and a reflectivity intensity signal; calculating an original reflectivity based on the reflectivity intensity signal, obtaining target distance information, and correcting the original reflectivity based on the target distance information to obtain a corrected reflectivity value; target motion features are generated through multi-frame time sequence analysis, the corrected reflectivity value, the target geometric features and the target motion features are fused based on a fusion strategy, and a target comprehensive feature vector is constructed; and inputting the target comprehensive feature vector into a classifier, and outputting an identification result of the tiny unmanned aerial vehicle.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of micro unmanned aerial vehicle detection and identification, and particularly relates to a micro unmanned aerial vehicle detection method and system based on target reflectivity. BACKGROUND

[0002] With the rapid development of unmanned aerial vehicle technology, especially the popularization of consumer and industrial micro unmanned aerial vehicle applications, the activities of such targets in the low-altitude field are increasingly frequent. Such targets usually have low radar cross section (RCS), small size, low flight altitude, and slow speed, etc., which pose a severe challenge to existing detection technologies. Traditional radar detection systems mainly rely on the strength and Doppler shift of the target echo signal for detection and tracking. However, due to the small physical size of micro unmanned aerial vehicles and the use of non-metallic materials, the echo signal is extremely weak and is easily disturbed by internal system noise and external environmental clutter, resulting in a significant decrease in detection probability and an increase in false alarm rate.

[0003] In addition, video target identification methods based on optical or infrared imaging have been widely introduced into the field of unmanned aerial vehicle detection in recent years. Such methods mainly identify targets by analyzing their geometric shape, motion pattern or trajectory characteristics, which to some extent makes up for the shortcomings of radar in low-altitude small target detection. However, existing video identification algorithms do not make full use of the electromagnetic reflection characteristics of the target itself, and fail to fully combine the reflection response characteristics of the target in different wave bands, so the identification accuracy is still not ideal in cases where the target and the background have low distinguishability, or there are similar appearance interference objects (such as birds, leaves, etc.).

[0004] On the other hand, complex environmental conditions (such as rain, fog, smoke, haze, etc.) will further exacerbate the detection difficulty. These meteorological conditions not only cause shading and scattering effects on optical imaging systems, reducing image quality, but also cause attenuation of radar waves, especially for Ka, W and other high-band radars, making it more difficult to effectively extract the weak unmanned aerial vehicle echo.

[0005] In summary, the existing detection means have obvious limitations when facing micro unmanned aerial vehicles with low observable characteristics: radar systems are limited by low signal-to-noise ratio and environmental interference, and video identification methods lack effective fusion of reflection characteristics, resulting in insufficient identification capability. Therefore, it is urgent to develop a new detection method that can comprehensively utilize target reflection characteristics and multi-modal sensing data to improve the reliable identification and stable tracking capability of micro unmanned aerial vehicles in complex environments. SUMMARY

[0006] The present application provides a micro unmanned aerial vehicle detection method and system based on target reflectivity, which solves the problem of low reflectivity targets being difficult to detect in complex environments, and realizes the identification of micro unmanned aerial vehicles by combining laser active detection and target reflectivity characteristic analysis.

[0007] In a first aspect, the present application provides a micro unmanned aerial vehicle detection method based on target reflectivity, comprising: emitting pulsed laser to a target area to cover a detection area through beam expansion and collimation; based on the pulsed laser, synchronously collecting target signals through double detectors designed in a common optical path to obtain target geometric features and reflectivity intensity signals; calculating original reflectivity based on the reflectivity intensity signals, obtaining target distance information, and correcting the original reflectivity based on the target distance information to obtain a corrected reflectivity value; generating target motion features through multi-frame time sequence analysis, and fusing the corrected reflectivity value, the target geometric features and the target motion features based on a fusion strategy to construct a target comprehensive feature vector; inputting the target comprehensive feature vector into a classifier to output an identification result of a micro unmanned aerial vehicle.

[0008] In combination with the first aspect, in a possible implementation manner, the based on the pulsed laser, synchronously collecting target signals through double detectors designed in a common optical path to obtain target geometric features and reflectivity intensity signals, comprises: the double detectors include an imaging detector and a non-imaging detector; the imaging detector is a CMOS detector, used for obtaining the target geometric features; the non-imaging detector is an APD avalanche photodiode array, used for obtaining the reflectivity intensity signals.

[0009] In combination with the first aspect, in a possible implementation manner, a field of view coincidence error of the imaging detector and the non-imaging detector is less than or equal to 0.1°, and a time synchronization error is less than or equal to 1 μs.

[0010] In combination with the first aspect, in a possible implementation manner, a calculation formula of the original reflectivity is: ; wherein, represents a peak value of echo light intensity in the reflectivity intensity signals; represents a peak value of emission light intensity in the reflectivity intensity signals; represents the original reflectivity.

[0011] In combination with the first aspect, in a possible implementation manner, the obtaining target distance information, and correcting the original reflectivity based on the target distance information to obtain a corrected reflectivity value, comprises: obtaining the distance information of the target through a TOF ranging technology ; According to the distance information, an atmospheric attenuation model is used to calculate a correction value The correction value is used to correct the original reflectivity to eliminate the influence of atmospheric attenuation, and a corrected reflectivity value is obtained.

[0012] In combination with the first aspect, in a possible implementation manner, the target motion feature is generated through multi-frame time sequence analysis, including: Based on the continuous multi-frame data of the imaging detector, the position coordinates of the target at adjacent time instants are obtained; The instantaneous speed and acceleration are calculated according to the time difference of the position coordinates; The trajectory direction angle is calculated according to the spatial relationship of the position coordinates; The position coordinates, instantaneous speed, acceleration and trajectory direction angle at adjacent time instants are taken as the target motion feature.

[0013] In combination with the first aspect, in a possible implementation manner, the fusion strategy includes: According to the size of the corrected reflectivity value, the weight of the target geometric feature in fusion is dynamically adjusted; wherein the dynamic adjustment includes: when the corrected reflectivity value is lower than 5%, the weight of the target geometric feature is increased; when the corrected reflectivity value is higher than 30%, the weight of the target geometric feature is reduced; A reflectivity-target geometric feature correlation factor is constructed; wherein the reflectivity-target geometric feature correlation factor includes a reflectivity-size correlation factor and a reflectivity-shape correlation factor; the reflectivity-target geometric feature correlation factor is expressed as: ; The reflectivity-shape correlation factor is expressed as: ; Wherein, represents the area of the light spot; represents the circularity; represents the correction value.

[0014] In combination with the first aspect, in a possible implementation manner, the corrected reflectivity value, the target geometric feature and the target motion feature are fused to construct a target comprehensive feature vector, including: According to the corrected reflectivity value, the target geometric feature and the target motion feature, a plurality of feature parameters are extracted; The plurality of parameters are respectively normalized, and the normalized plurality of feature parameters are spliced in a fixed order to obtain the target comprehensive feature vector.

[0015] In a possible implementation manner of the first aspect, the classifier is a classifier based on a machine learning model, and at least one of a support vector machine and a convolutional neural network.

[0016] In a second aspect, the present application provides a micro unmanned aerial vehicle detection system based on target reflectivity, comprising: a laser emission module for emitting pulsed laser to a target area, covering the detection area after beam expansion and collimation; A dual-detector module is configured to synchronously collect target signals based on the pulsed laser through a dual-detector in common optical path design, to obtain target geometric features and reflectivity intensity signals; A signal processing module is configured to calculate original reflectivity based on the reflectivity intensity signals, to obtain target distance information, and to correct the original reflectivity based on the target distance information to obtain a corrected reflectivity value; A feature fusion module is configured to generate target motion features through multi-frame time sequence analysis, and to fuse the corrected reflectivity value, the target geometric features and the target motion features based on a fusion strategy to construct a target comprehensive feature vector; A classification and recognition module is configured to input the target comprehensive feature vector into a classifier to output a recognition result of a micro unmanned aerial vehicle.

[0017] The one or more technical solutions provided in the present application have at least the following technical effects or advantages: In the present application, pulsed laser is emitted to a target area, and the laser uniformly covers the entire detection area after beam expansion and collimation, effectively increasing the irradiation range and improving the signal-to-noise ratio. Based on the pulsed laser, target signals are synchronously collected through a dual-detector in common optical path design to synchronously obtain target geometric features and reflectivity intensity signals, and original reflectivity is calculated accordingly. The common optical path design ensures that the two types of signals completely correspond in time and space, reducing calibration errors. Target distance information is obtained, and the distance information is used to correct the original reflectivity to obtain a more accurate corrected reflectivity value, thereby eliminating the influence of distance factors on reflectivity intensity and improving the comparability and reliability of reflectivity features. Target motion features are generated through multi-frame time sequence analysis, and the corrected reflectivity value, target geometric features and generated motion features are fused based on a fusion strategy to construct a comprehensive feature vector that can comprehensively describe target properties, enhancing feature expression capability. The target comprehensive feature vector is input into a classifier, and a recognition result of a micro unmanned aerial vehicle is output based on multi-dimensional fusion features, improving the recognition accuracy and robustness of low observable targets in complex environments. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 A micro unmanned aerial vehicle detection method based on target reflectivity is provided for the embodiments of the present application. Figure 2A target comprehensive feature vector processing flowchart provided by the embodiment of the present application is provided. Figure 3 A schematic diagram of a micro unmanned aerial vehicle detection system based on target reflectivity provided by the embodiment of the present application is provided. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the protection scope of the present application.

[0020] The present application provides a micro unmanned aerial vehicle detection method based on target reflectivity, referring to Figure 1 The method comprises the following steps S101 to S104.

[0021] S101, emitting pulsed laser to a target area, covering a detection area through beam expansion and collimation; Illustratively, the wavelength of the pulsed laser can be selected as near-infrared or long-wave infrared.

[0022] S102, based on the pulsed laser, synchronously collecting target signals through double detectors designed in a common optical path to obtain target geometric features and reflectivity intensity signals; Specifically, in step S102, based on the pulsed laser, synchronously collecting target signals through double detectors designed in a common optical path to obtain target geometric features and reflectivity intensity signals, comprising: the double detectors comprising an imaging detector and a non-imaging detector; the imaging detector is a CMOS detector, used for acquiring target geometric features; the non-imaging detector is an APD avalanche photodiode array, used for acquiring reflectivity intensity signals.

[0023] Here, the target geometric features include: spot area (target size), circularity (target shape regularity), aspect ratio (target contour feature).

[0024] Here, the field of view coincidence error of the imaging detector and the non-imaging detector is less than or equal to 0.1°, and the time synchronization error is less than or equal to 1μs.

[0025] S103, calculating original reflectivity based on the reflectivity intensity signals, acquiring target distance information, and correcting the original reflectivity based on the target distance information to obtain a corrected reflectivity value; Specifically, in step S103, the target distance information is acquired, and the original reflectivity is corrected based on the target distance information to obtain a corrected reflectivity value, comprising the following steps S1031 to S1032.

[0026] S1031, obtaining distance information of the target by TOF ranging technology S1032, using an atmospheric attenuation model to calculate a correction value according to the distance information to correct the original reflectivity to eliminate the influence of atmospheric attenuation, and obtain a corrected reflectivity value.

[0027] Here, the calculation formula of the original reflectivity is: Among them, represents the peak value of the echo light intensity in the reflectivity intensity signal; represents the peak value of the transmitted light intensity in the reflectivity intensity signal; represents the original reflectivity.

[0028] S104, generating target motion features through multi-frame time sequence analysis, and fusing the corrected reflectivity value, target geometric features and target motion features based on a fusion strategy to construct a target comprehensive feature vector; Specifically, in step S104, target motion features are generated through multi-frame time sequence analysis, including the following steps S1041 to S1044.

[0029] S1041, based on the continuous multi-frame data of the imaging detector, obtaining the position coordinates of the target at adjacent time points; S1042, calculating the instantaneous speed and acceleration according to the time difference of the position coordinates; S1043, calculating the trajectory direction angle according to the spatial relationship of the position coordinates; S1044, taking the position coordinates, instantaneous speed, acceleration and trajectory direction angle at adjacent time points as target motion features.

[0030] Here, the fusion strategy includes: (1) dynamically adjusting the weight of the target geometric features in the fusion according to the size of the corrected reflectivity value; wherein the dynamic adjustment includes: when the corrected reflectivity value is lower than 5%, increasing the weight of the target geometric features; when the corrected reflectivity value is higher than 30%, reducing the weight of the target geometric features; (2) constructing a reflectivity-target geometric feature correlation factor; wherein the reflectivity-target geometric feature correlation factor includes a reflectivity-size correlation factor and a reflectivity-shape correlation factor; the reflectivity-target geometric feature correlation factor is represented as: The reflectivity-shape correlation factor is represented as: Among them, represents the spot area;​​​​ Indicates roundness; This indicates a correction value.

[0031] Specifically, in step S104, the corrected reflectivity value, target geometric features, and target motion features are fused to construct a comprehensive target feature vector, including: (1) Extract multiple feature parameters based on the corrected reflectivity value, target geometric features, and target motion features; (2) Normalize the multiple parameters respectively, and then concatenate the normalized feature parameters in a fixed order to obtain the target comprehensive feature vector.

[0032] See Figure 2 This is a flowchart of the process for processing the target comprehensive feature vector.

[0033] For example, reflectivity features: taking the corrected true reflectivity Rcorrected (single value, range 0~100%), and the rate of change of reflectivity over 3 consecutive frames, expressed by the formula: This reflects the stability of the target material.

[0034] The geometric features of the target include: target size and shape features.

[0035] Target size: Spot area S (number of pixels), equivalent diameter The diameter of the light spot is approximated as a circle; shape characteristics: circularity. ;in, The circumference of the light spot is represented by a value between 0 and 1, with values ​​closer to 1 indicating a circular shape; the aspect ratio of the circumscribed rectangle is also considered. Width / height, distinguishing between elongated and square targets.

[0036] Motion characteristics: instantaneous velocity, acceleration, and trajectory direction angle. Instantaneous velocity is expressed as: ; in, The coordinates of the target center; Acceleration is expressed as: This reflects whether the motion is uniform; The trajectory direction angle is expressed as: , and the angle between it and the horizontal direction.

[0037] The above 10 characteristic parameters are normalized to eliminate dimensional differences: ; in, Represents the original eigenvalues; This represents the standardized value.

[0038] The 10 features after normalization are spliced in a fixed order to form a 10-dimensional target comprehensive feature vector, and the target comprehensive feature vector is represented as: ; In S105, the target comprehensive feature vector is input into a classifier to output the identification result of the micro unmanned aerial vehicle.

[0039] Specifically, in step S105, the classifier is a classifier based on a machine learning model, including at least one of a support vector machine and a convolutional neural network.

[0040] For example, the support vector machine SVM and the convolutional neural network CNN based on the machine learning model are used to build a data set to distinguish target types. The introduction of the target reflectivity parameter corresponding to the laser wave band on the basis of the traditional algorithm of target motion trajectory and geometric shape can greatly increase the distinction of different types of targets such as stealth devices, birds, unmanned aerial vehicles in a weak target identification environment.

[0041] In a second aspect, the present application provides a micro unmanned aerial vehicle detection system based on target reflectivity, referring to Figure 3 The system comprises: A laser emission module for emitting pulsed laser to a target area, covering the detection area through beam expansion and collimation; A double detector module for synchronously collecting target signals through double detectors designed in a common optical path based on pulsed laser to obtain target geometric features and reflectivity intensity signals; A signal processing module for calculating original reflectivity based on the reflectivity intensity signal, obtaining target distance information, and correcting the original reflectivity based on the target distance information to obtain a corrected reflectivity value; A feature fusion module for generating target motion features through multi-frame time sequence analysis, and fusing the corrected reflectivity value, target geometric features and target motion features based on a fusion strategy to construct a target comprehensive feature vector; A classification and identification module for inputting the target comprehensive feature vector into a classifier to output the identification result of the micro unmanned aerial vehicle.

[0042] The present application combines target optical reflectivity with target geometric features, and realizes deep fusion through modeling of the correlation between physical characteristics and morphological features. This is specifically embodied in the following two aspects: 1. Feature screening layer: dynamic weighting of target geometric features based on reflectivity.

[0043] For low reflectivity targets (R<5%), the target geometric features (such as spot area, shape) are easily disturbed by noise, so the weight of the target geometric features in the feature vector is increased to compensate for the weakness of the reflectivity signal; For high reflectivity targets (R>30%), the reflectivity feature stability is high, the target geometric feature weight is reduced, and redundant information interference is avoided.

[0044] Here, in daily life, the core of the division between high optical reflectivity targets and low optical reflectivity targets is the reflection ability (reflectivity is usually expressed in percentage, high reflectivity is generally > 50%, and low reflectivity is generally < 20%) of objects to visible light (or specific band light). There are significant differences in material, appearance, application scenario and optical characteristics between the two.

[0045] The core characteristics of high optical reflectivity objects are: visual characteristics: under natural light or light, the surface brightness of the object is high, and "high light area" is easy to form. Some specular reflection targets can be clearly imaged (such as mirrors can reflect images).

[0046] Optical characteristics: reflectivity is usually between 50%-100% (such as mirror glass reflectivity is about 80%-90%, white latex paint is about 70%-85%), and light absorption and transmission are very low (except for transparent high-reflectivity materials such as some optical glass).

[0047] Material commonality: mostly smooth surface (specular reflection) or high whiteness material (diffuse reflection), the absorption coefficient of the material itself to visible light is low (such as the free electrons of metal are easy to reflect light, and the particles of white pigment are easy to scatter light).

[0048] The core characteristics of low optical reflectivity objects are: visual characteristics: the surface color is mostly dark black, dark brown and dark gray, and there is no obvious reflection after light irradiation. It is visually "dark", and even can absorb ambient stray light (such as photo studio light-absorbing cloth can eliminate background reflection).

[0049] Optical characteristics: reflectivity is usually between 2%-20% (such as pure black paint reflectivity is about 2%-5%, and frosted black plastic is about 10%-15%), most of the light is absorbed by the object (converted into heat or other energy), and the transmission rate is very low (except for black transparent materials such as sunglasses lenses).

[0050] Material commonality: mostly dark pigment coating, rough surface material or high absorption material (such as the carbon atom structure of graphite is easy to absorb visible light), the surface has no smooth mirror, and the light irradiation occurs "weak diffuse reflection" or is directly absorbed.

[0051] 2. Feature interaction layer: Construct reflectivity-target geometric feature correlation factor. Add 2 cross-features to quantify the internal correlation between reflectivity and target geometric features: reflectivity-size correlation factor, under the same size, the reflectivity of the UAV is usually higher than that of the flying bird (the UAV is mostly metal / plastic, and the flying bird is feather), which can amplify the difference); reflectivity-shape correlation factor, the shape of the UAV is more regular (C is larger, 1-C is smaller), while the shape of the flying bird is irregular due to the flapping of the wings (C is smaller, 1-C is larger), which can distinguish the two in combination with reflectivity.

[0052] In the above manner, reflectivity (reflecting the essence of the target material) and target geometric features (reflecting the shape of the target) form a complement: reflectivity can distinguish "same shape different material" targets (such as plastic drones and bionic birds), target geometric features can distinguish "same material different shape" targets (such as multi-rotor drones and fixed-wing drones), and the combination of the two improves the identification accuracy of micro drones (compared to a single feature).

[0053] In one specific embodiment provided by the present application, 1. Hardware deployment: laser (wavelength 1550 nm, pulse energy 10 mJ, pulse width 10 ns). Dual-probe common optical path system (imaging probe: 1920x1080 CMOS; non-imaging probe: APD avalanche photodiode array).

[0054] 2. Software processing: after the echo signal is sampled by ADC, it is denoised by adaptive filtering (Wiener filtering). Based on the YOLOv9 model, the target classifier is trained, and the recognition accuracy is above 95%. Dynamic adjustment: when the target reflectivity is lower than 1%, automatically switch to long pulse mode (pulse width 50 ns) to improve the signal-to-noise ratio.

[0055] Reflectivity + feature fusion mechanism: for the first time, the target reflectivity is combined with the target geometric features to improve the identification accuracy of micro drones. Adaptive laser parameter adjustment: dynamically optimize the detection parameters according to the reflectivity and the environment to expand the application range of the system. Dual-probe cooperative processing: imaging and non-imaging probes are complementary, taking into account resolution and real-time performance.

[0056] The CMOS probe (1024x1024 resolution) collects imaging information of the reflected laser of the target and directly outputs image data containing the spatial shape of the target. From the image, target geometric features can be extracted, such as spot area (target size), circularity (target shape regularity), aspect ratio (target contour feature), etc., providing morphological basis for feature fusion.

[0057] The APD array, as a non-imaging probe, focuses on capturing the peak signal of the echo light intensity , combined with the emission light intensity of the laser , to calculate the original reflectivity by formula.

[0058] The high sensitivity characteristic (avalanche gain effect) can amplify weak echo signals, and is particularly suitable for precise measurement of the reflectivity of low reflectivity targets (such as small unmanned aerial vehicles with reflectivity <1%), and provides reliable raw data for subsequent correction of reflectivity.

[0059] The dual probe adopts a common optical path design, ensures that the detection fields of view coincide and the time synchronization error is ≤1μs, realizes the space-time matching of “target geometric features-reflectivity”, and provides hardware-level collaborative support for the innovation point of “reflectivity + target geometric feature fusion”, and takes into account the imaging resolution and reflectivity measurement accuracy.

[0060] The various embodiments in the specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the difference from other embodiments. The whole or part of the present application can be used in a plurality of general or special computer system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, mobile communication terminals, multi-processor systems, microprocessor-based systems, programmable electronic devices, network PCs, small computers, large computers, distributed computing environments including any of the above systems or devices, and the like.

[0061] The above examples are only used to illustrate the technical solutions of the present application, and are not limited to the present application; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can still be modified, or some or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the present application.

Claims

1. A method for detecting a micro unmanned aerial vehicle based on target reflectivity, characterized in that, The method comprises the following steps: Pulse laser is emitted to a target area, and a probe area is covered by beam expansion and collimation; Based on the pulse laser, a target signal is synchronously collected by double detectors in a common optical path design to obtain a target geometric feature and a reflectivity intensity signal; Based on the reflectivity intensity signal, an original reflectivity is calculated, target distance information is obtained, and the original reflectivity is corrected based on the target distance information to obtain a corrected reflectivity value; Target motion features are generated by multi-frame time sequence analysis, and the corrected reflectivity value, the target geometric feature and the target motion feature are fused based on a fusion strategy to construct a target comprehensive feature vector; The target comprehensive feature vector is input into a classifier to output an identification result of a micro unmanned aerial vehicle. 2.The target reflectivity-based micro unmanned aerial vehicle detection method of claim 1, wherein, The method comprises the following steps: The double detectors comprise an imaging detector and a non-imaging detector; The imaging detector is a CMOS detector, and is used to obtain the target geometric feature; The non-imaging detector is an APD avalanche photodiode array, and is used to obtain the reflectivity intensity signal. 3.The target reflectivity-based micro unmanned aerial vehicle detection method of claim 2, wherein, The field of view coincidence error of the imaging detector and the non-imaging detector is less than or equal to 0.1°, and the time synchronization error is less than or equal to 1 μs. 4.The method of claim 1, wherein, The calculation formula of the original reflectivity is as follows: ； wherein, represents a peak value of the intensity of the reflected light in the reflectance intensity signal; represents a peak value of the intensity of the emitted light in the reflectance intensity signal; represents the raw reflectance. 5.The target reflectivity-based micro unmanned aerial vehicle detection method of claim 1, wherein, The method comprises the following steps: acquiring distance information of the target by TOF ranging technology ; According to the distance information, an atmospheric attenuation model is used to calculate a correction value The original reflectivity is corrected by the correction value to eliminate the influence of atmospheric attenuation, and a corrected reflectivity value is obtained. 6.The target reflectivity-based micro unmanned aerial vehicle detection method of claim 1, wherein, The method comprises the following steps: Based on continuous multi-frame data of the imaging detector, position coordinates of the target at adjacent time instants are obtained; Instantaneous speed and acceleration are calculated based on the time difference of the position coordinates; A trajectory direction angle is calculated based on the spatial relationship of the position coordinates; The position coordinates, the instantaneous speed, the acceleration and the trajectory direction angle at adjacent time instants are taken as the target motion features. 7.The target reflectivity-based micro unmanned aerial vehicle detection method of claim 1, wherein, The fusion strategy comprises the following steps: The weight of the target geometric feature in fusion is dynamically adjusted according to the size of the corrected reflectivity value; wherein the dynamic adjustment comprises: when the corrected reflectivity value is lower than 5%, the weight of the target geometric feature is increased; and when the corrected reflectivity value is higher than 30%, the weight of the target geometric feature is decreased; A reflectivity-target geometric feature correlation factor is constructed; wherein the reflectivity-target geometric feature correlation factor comprises a reflectivity-size correlation factor and a reflectivity-shape correlation factor; and the reflectivity-target geometric feature correlation factor is expressed as: ； The reflectivity-shape correlation factor is expressed as: ； wherein, represents a spot area; represents a circularity; represents a correction value. 8.The target reflectivity-based micro unmanned aerial vehicle detection method of claim 1, wherein, The method comprises the following steps: A plurality of feature parameters are extracted according to the corrected reflectivity value, the target geometric feature and the target motion feature; The plurality of feature parameters are normalized respectively, and the normalized plurality of feature parameters are spliced in a fixed order to obtain the target comprehensive feature vector. 9.The target reflectivity-based micro unmanned aerial vehicle detection method of claim 1, wherein, The classifier is a classifier based on a machine learning model, and comprises at least one of a support vector machine and a convolutional neural network.

10. A micro unmanned aerial vehicle detection system based on target reflectivity, characterized by, The method comprises the following steps: The laser emission module is configured to emit pulsed laser to a target area, and the expanded and collimated laser covers a detection area; The dual-detector module is configured to synchronously collect target signals through dual detectors in a common optical path based on the pulsed laser, and obtain target geometric features and reflectivity intensity signals; The signal processing module is configured to calculate original reflectivity based on the reflectivity intensity signals, obtain target distance information, correct the original reflectivity based on the target distance information, and obtain a corrected reflectivity value; The feature fusion module is configured to generate target motion features through multi-frame time sequence analysis, fuse the corrected reflectivity value, the target geometric features and the target motion features based on a fusion strategy, and construct a target comprehensive feature vector; The classification and recognition module is configured to input the target comprehensive feature vector into a classifier, and output an identification result of a micro unmanned aerial vehicle.

Images