Target feature recognition method and device based on back propagation neural network and orbital angular momentum feature spectrum

The method and device use a back propagation neural network and orbital angular momentum characteristic spectrum to correct and identify target shapes in atmospheric turbulence, addressing the inefficiencies of existing laser target recognition methods.

CN119087394APending Publication Date: 2024-12-06HARBIN INST OF TECH
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202411137596.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-19
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

Existing laser target recognition methods fail to achieve efficient and accurate identification of typical features in atmospheric turbulence due to interference from the atmosphere.

Method used

A method and device using a back propagation neural network and orbital angular momentum characteristic spectrum to identify the target shape of a vortex beam generating unit, a vortex beam is irradiated onto a typical object in an atmospheric turbulence environment, followed by phase distortion correction and target angular spectrum analysis using a spatial light modulator and optical power detector, and finally identified using a back propagation neural network.

Benefits of technology

The method effectively recognizes targets with typical shapes in atmospheric turbulence by combining phase compensation with intelligent algorithm recognition, achieving accurate target identification.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119087394A_ABST
    Figure CN119087394A_ABST
Patent Text Reader

Abstract

The invention discloses a target feature recognition method and device based on a back propagation neural network and an orbital angular momentum feature spectrum, belongs to the field of laser target recognition, and aims to solve the problem that a typical feature target cannot be efficiently recognized in atmospheric turbulence in an existing laser target recognition method technology. The method comprises the following steps: step 1, irradiating a vortex light beam on a typical object in an atmospheric turbulence environment through a transmitting system and generating an echo signal; step 2, performing phase distortion correction on the echo signal to obtain a corrected light beam; step 3, acquiring a target angular spectrum of the correction light beam; and step 4, inputting the target angular spectrum into a back propagation neural network so as to identify the target shape of the typical object. The device comprises a vortex beam generation unit, a 4F system, an emission optical system, a first atmospheric turbulence simulator, a first beam splitter, a second atmospheric turbulence simulator, a receiving system, a distortion correction module, an angular spectrum analysis module and an algorithm identification module. A target shape is identified by an algorithm identification module.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a target feature recognition method and device based on a back-propagation neural network and an orbital angular momentum characteristic spectrum, and belongs to the field of laser target recognition. Background Art

[0002] Laser target recognition can identify typical features of targets and has broad application potential in industrial control, military industry, aerospace, etc. However, there is currently a lack of effective means to identify typical features of actual targets. Atmospheric turbulence will seriously interfere with the transmission of light beams. Currently, it is impossible to achieve efficient and accurate recognition of typical feature targets in atmospheric turbulence. Summary of the invention

[0003] Aiming at the problem that the existing laser target recognition method technology cannot realize efficient recognition of typical characteristic targets in atmospheric turbulence, the present invention provides a target characteristic recognition method and device based on back propagation neural network and orbital angular momentum characteristic spectrum.

[0004] In one aspect of the present invention, a method for identifying target features based on a back propagation neural network and an orbital angular momentum characteristic spectrum is provided, the method comprising the following steps:

[0005] Step 1: The vortex beam is irradiated onto a typical object in an atmospheric turbulence environment through a transmitting system to generate an echo signal;

[0006] Step 2: performing phase distortion correction on the echo signal to obtain a correction light beam;

[0007] Step 3, obtaining the target angular spectrum of the correction beam;

[0008] Step 4: Input the target angular spectrum into the back propagation neural network to identify the target shape of a typical object.

[0009] Preferably, the phase distortion correction process in step 2 is:

[0010] Step 21: Using a wavefront sensor to obtain the distorted light beam light field;

[0011] Step 22: input the distorted light beam field and the original light beam field into the feedback signal controller together, and use the GS algorithm to obtain the original light beam phase in the feedback signal control;

[0012] Step 23: Use the reverse phase of the original light beam to control the phase loading of the spatial light modulator, thereby correcting the distorted light beam and outputting the corrected light beam.

[0013] Preferably, the process of obtaining the target angular spectrum of the correction light beam in step 3 is:

[0014] The spatial light modulator is used to demodulate the correction beam light field, and then the optical power detector is used to detect the power of each orbital angular momentum mode, and finally the orbital angular momentum spiral spectrum of the correction beam light field is obtained as the target angular spectrum.

[0015] Preferably, the training back propagation neural network in step 4 comprises an input layer, a hidden layer, and an output layer, the input signal of the back propagation neural network is the target angular spectrum of the echo signal of a typical object in an atmospheric turbulence environment, and the output signal is the shape of the typical object; the training process is:

[0016] Step 4: Determine the number of layers of the network, the number of neurons in each hidden layer, and randomly assign values ​​to the weights and thresholds of the network;

[0017] Step 42: Input a training sample, select an activation function, calculate the network, and obtain the output value of the network;

[0018] Step 43: Calculate the error between the output value and the expected value; if the error meets the requirement, stop training the network; if the error does not meet the requirement, execute step 44;

[0019] Step 44: Back propagate and adjust the network weights and thresholds; then, input the next training sample and return to step 42 for a loop calculation.

[0020] Preferably, the atmospheric turbulence environment is simulated by the phase screen method, and the atmospheric turbulence process is replaced by a random phase screen set with equal spacing Δz and simulated by the power spectrum inversion method. The light beam is freely transmitted between the random phase screens according to the angular spectrum diffraction theory. The modified VonKarman spectrum is used, and the phase screen can be expressed as:

[0021]

[0022] Where (x, y) is the coordinate of the phase screen xoy plane, x = jΔx, y = lΔy, Δx, Δy are the sampling intervals of the x-axis and y-axis, j, l are integers, and h(κ x ,κ y ) is a Gaussian random number with unit variance and zero mean;

[0023] Φ φ (f) is the phase power spectrum density of the phase screen, which can be obtained by the following formula:

[0024]

[0025] Where, f is the spatial frequency, f = κ / 2π, κ represents the three-dimensional space wave number;

[0026] L 0 is the outer scale of turbulence; l 0 is the inner scale of turbulence;

[0027] r 0 is the atmospheric coherence length, is the turbulence intensity, z is the z-axis coordinate of the phase screen rectangular coordinate system, and Δz is the z-axis sampling interval.

[0028] In another aspect of the present invention, a target feature recognition device based on a back propagation neural network and an orbital angular momentum characteristic spectrum is provided, the device is used to implement the method described in any one of claims 1 to 5, the device includes a vortex beam generating unit, a 4F system 4, a transmitting optical system 5, a No. 1 atmospheric turbulence simulator 6, a No. 1 beam splitter 7, a No. 2 atmospheric turbulence simulator 9, a receiving system 10, a distortion correction module 11, an angular spectrum analysis module 12 and an algorithm recognition module 13;

[0029] The vortex beam generating unit outputs a single-order vortex beam, and the 4F system 4 selects the first diffraction order of the single-order vortex beam and transmits it through the transmitting optical system 5. It passes through the No. 1 atmospheric turbulence simulator 6 and then the No. 1 beam splitter 7 to irradiate the typical object 8 to generate an echo signal. The receiving system 10 receives the echo signal passing through the No. 2 atmospheric turbulence simulator 9, and the phase of the distorted beam in the echo signal is corrected by the distortion correction module 11. The target angular spectrum is obtained by the angular spectrum analysis module 12, and the target shape is identified by the algorithm recognition module 13.

[0030] Preferably, the vortex beam generating unit includes a laser 1, a single-mode optical fiber 2 and a spatial light modulator No. 1 3. The laser 1 emits laser light to the single-mode optical fiber 2, the single-mode optical fiber 2 collimates the light beam, and then the spatial light modulator No. 1 3 performs orbital angular momentum quantum control on the light beam to generate a single-order vortex beam signal.

[0031] Preferably, the distortion correction module 11 includes a No. 2 spatial light modulator 11-1, a No. 2 beam splitter 11-2, a wavefront sensor 11-3 and a feedback signal controller 11-4; the distorted light beam is divided into two beams by the No. 2 beam splitter 11-2, one of which is detected by the wavefront sensor 11-3 to output the distorted light beam light field, which is input to the feedback signal controller 11-4 together with the original light beam light field, the feedback signal controller 11-4 outputs the original light beam phase, and loads the reverse phase of the original light beam to the No. 2 spatial light modulator 11-1, and the other beam output by the No. 2 beam splitter 11-2 is subjected to phase distortion correction via the No. 2 spatial light modulator 11-1 to output a corrected light beam.

[0032] Preferably, the angular spectrum analysis module 12 includes a No. 3 spatial light modulator and an optical power detector. The correction light beam is input into the No. 3 spatial light modulator for demodulation of the light field. The demodulated light beam is input into the optical power detector. According to the power of each orbital angular momentum mode detected by the optical power detector, the orbital angular momentum spiral spectrum of the light field is obtained as the target angular spectrum.

[0033] Preferably, the algorithm recognition module 13 is implemented by a back-propagation neural network, and the characteristic shapes of typical objects are recognized through the orbital angular momentum spiral spectrum of the light field. The characteristic shapes of the typical objects include triangles, squares and regular pentagons.

[0034] Beneficial effects of the present invention: The present invention introduces the GS algorithm to restore and correct the distorted light beam carrying the target shape feature information, and selects the back propagation neural network to identify typical targets of different shapes. The present invention combines phase compensation with intelligent algorithm recognition, which can effectively realize the recognition of targets with typical characteristic shapes in atmospheric turbulence. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a block diagram of the target feature recognition method based on back propagation neural network and orbital angular momentum characteristic spectrum of the present invention;

[0036] Figure 2 It is a structural diagram of the distortion correction module;

[0037] Figure 3 It is a schematic diagram of the atmospheric turbulence phase screen loaded by the atmospheric turbulence simulator of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0040] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0041] Specific implementation method 1: The following is combined Figures 1 to 3 The present embodiment is described. The target feature recognition method based on the back propagation neural network and the orbital angular momentum characteristic spectrum described in the present embodiment comprises the following steps:

[0042] Step 1: The vortex beam is irradiated onto a typical object in an atmospheric turbulence environment through a transmitting system to generate an echo signal;

[0043] Step 2: performing phase distortion correction on the echo signal to obtain a correction light beam;

[0044] Step 3, obtaining the target angular spectrum of the correction beam;

[0045] Step 4: Input the target angular spectrum into the back propagation neural network to identify the target shape of a typical object.

[0046] The atmospheric turbulence environment is simulated by the phase screen method. The atmospheric turbulence process is replaced by a random phase screen set with equal spacing Δz and simulated by the power spectrum inversion method. The light beam is freely transmitted between the random phase screens according to the angular spectrum diffraction theory. Using the modified VonKarman spectrum, the phase screen can be expressed as:

[0047]

[0048] Where (x, y) is the coordinate of the phase screen xoy plane, x = jΔx, y = lΔy, Δx, Δy are the sampling intervals of the x-axis and y-axis, j, l are integers, and h(κ x ,κ y ) is a Gaussian random number with unit variance and zero mean;

[0049] Φ φ (f) is the phase power spectrum density of the phase screen, which can be obtained by the following formula:

[0050]

[0051] Where, f is the spatial frequency, f = κ / 2π, κ represents the three-dimensional space wave number;

[0052] L 0 is the outer scale of turbulence; l 0 is the inner scale of turbulence;

[0053] r 0 is the atmospheric coherence length, is the turbulence intensity, z is the z-axis coordinate of the phase screen rectangular coordinate system, and Δz is the z-axis sampling interval.

[0054] The phase distortion correction process in step 2 is:

[0055] Step 21: Using a wavefront sensor to obtain the distorted light beam light field;

[0056] Step 22: input the distorted light beam field and the original light beam field into the feedback signal controller together, and use the GS algorithm to obtain the original light beam phase in the feedback signal control;

[0057] Step 23: Use the reverse phase of the original light beam to control the phase loading of the spatial light modulator, thereby correcting the distorted light beam and outputting the corrected light beam.

[0058] The GS algorithm obtains the phase distribution of the initial light beam through continuous diffraction iteration through the light field distribution of the initial light beam and the distorted light beam after atmospheric turbulence, so that the initial light beam becomes a distorted light beam after one diffraction. This means that the obtained phase has the same effect as the atmospheric turbulence, and the reverse phase is used as the initial pre-correction screen, thereby pre-correcting the distorted light beam and offsetting the influence of atmospheric turbulence.

[0059] The process of obtaining the target angular spectrum of the correction beam in step 3 is:

[0060] The spatial light modulator is used to demodulate the correction beam light field, and then the optical power detector is used to detect the power of each orbital angular momentum mode, and finally the orbital angular momentum spiral spectrum of the correction beam light field is obtained as the target angular spectrum.

[0061] In step 4, the training back propagation neural network contains an input layer, a hidden layer, and an output layer. The input signal of the back propagation neural network is the target angular spectrum of the echo signal of a typical object in an atmospheric turbulence environment, and the output signal is the shape of the typical object. The training process is:

[0062] Step 4: Determine the number of layers of the network, the number of neurons in each hidden layer, and randomly assign values ​​to the weights and thresholds of the network;

[0063] Step 42: Input a training sample, select an activation function, calculate the network, and obtain the output value of the network;

[0064] Step 43: Calculate the error between the output value and the expected value; if the error meets the requirement, stop training the network; if the error does not meet the requirement, execute step 44;

[0065] Step 44: Back propagate and adjust the network weights and thresholds; then, input the next training sample and return to step 42 for a loop calculation.

[0066] Specific implementation method 2: The following is combined Figures 1 to 3 This embodiment is described. The target feature recognition based on the back propagation neural network and the orbital angular momentum characteristic spectrum described in this embodiment is used in the target feature recognition method based on the back propagation neural network and the orbital angular momentum characteristic spectrum described in the first embodiment. The unique phase characteristics of the vortex beam are used. Since there are differences in the orbital angular momentum spectra corresponding to targets of different shapes, this difference can be used to identify characteristic targets. The GS algorithm is used to optimize and compensate the vortex beam phase to further improve the correction of the distorted vortex beam that has passed through atmospheric turbulence, and then the back propagation neural network is selected to identify targets of different shapes with high accuracy.

[0067] The target feature recognition device based on the back propagation neural network and the orbital angular momentum characteristic spectrum includes a vortex beam generating unit, a 4F system 4, a transmitting optical system 5, a No. 1 atmospheric turbulence simulator 6, a No. 1 beam splitter 7, a No. 2 atmospheric turbulence simulator 9, a receiving system 10, a distortion correction module 11, an angular spectrum analysis module 12 and an algorithm recognition module 13;

[0068] The vortex beam generating unit outputs a single-order vortex beam, and the 4F system 4 selects the first diffraction order of the single-order vortex beam and transmits it through the transmitting optical system 5. It passes through the No. 1 atmospheric turbulence simulator 6 and then the No. 1 beam splitter 7 to irradiate the typical object 8 to generate an echo signal. The receiving system 10 receives the echo signal passing through the No. 2 atmospheric turbulence simulator 9, and the phase of the distorted beam in the echo signal is corrected by the distortion correction module 11. The target angular spectrum is obtained by the angular spectrum analysis module 12, and the target shape is identified by the algorithm recognition module 13.

[0069] The vortex beam generating unit includes a laser 1, a single-mode optical fiber 2 and a spatial light modulator No. 1 3. The laser 1 emits laser light to the single-mode optical fiber 2, the single-mode optical fiber 2 collimates the light beam, and then the spatial light modulator No. 1 3 performs orbital angular momentum quantum control on the light beam to generate a single-order vortex beam signal.

[0070] The first spatial light modulator 3 is a polarization-dependent liquid crystal reflective phase modulator, which can modulate the horizontal X direction of the light field; the modulation process of the Gaussian mode laser signal requires loading the phase diagram of the vortex light onto the spatial light modulator, and the vortex light beam can be generated by laser incidence;

[0071] The 4F system 4 selects a first-order modulation signal from the incident modulation signal and transmits it to the emission optical system 5;

[0072] The No. 1 atmospheric turbulence simulator 6 and the No. 2 atmospheric turbulence simulator 9 are to load a random phase screen simulated by the power spectrum inversion method on the transmission type spatial light modulator to replace the atmospheric turbulence process, and the light beam is distorted when passing through the transmission type spatial light modulator;

[0073] The typical targets 8 are typical targets with different shape features, such as an equilateral triangle, a square, and a regular pentagon;

[0074] The receiving system 10 receives the echo signal, corrects the distorted light beam through the distortion correction module 11, obtains the target angular spectrum through the angular spectrum analysis module 12, and the target angular spectrum is the characteristic spectrum. The algorithm recognition module 13 recognizes the target shape.

[0075] The distortion correction module 11 includes a second spatial light modulator 11-1, a second beam splitter 11-2, a wavefront sensor 11-3 and a feedback signal controller 11-4; the distorted light beam is divided into two beams by the second beam splitter 11-2, one of which is detected by the wavefront sensor 11-3 to output the distorted light beam field, and is input to the feedback signal controller 11-4 together with the original light beam field, the feedback signal controller 11-4 outputs the original light beam phase, and loads the reverse phase of the original light beam to the second spatial light modulator 11-1, and the other beam output by the second beam splitter 11-2 is subjected to phase distortion correction by the second spatial light modulator 11-1 to output the corrected light beam. The feedback signal controller 11-4 calculates and generates a pre-compensated phase screen in combination with the GS phase recovery algorithm. The GS algorithm obtains the phase distribution of the initial light beam through continuous diffraction iteration through the light field distribution of the initial light beam and the distorted light beam after atmospheric turbulence, so that the initial light beam becomes a distorted light beam after one diffraction. This means that the obtained phase has the same effect as the atmospheric turbulence, and the reverse phase is used as the initial pre-correction screen, thereby pre-correcting the distorted light beam and offsetting the influence of atmospheric turbulence.

[0076] The angular spectrum analysis module 12 includes a No. 3 spatial light modulator and an optical power detector. The correction light beam is input to the No. 3 spatial light modulator for demodulation of the light field. The demodulated light beam is input to the optical power detector. According to the power of each orbital angular momentum mode detected by the optical power detector, the orbital angular momentum spiral spectrum of the light field is obtained as the target angular spectrum.

[0077] The algorithm recognition module 13 is implemented by a back propagation neural network, and recognizes characteristic shapes of typical objects through the orbital angular momentum spiral spectrum of the light field. The characteristic shapes of the typical objects include triangles, squares and regular pentagons.

[0078] The basic principle of the reverse neural network is: the reverse neural network is a supervised multi-layer feedforward neural network. The input signal is forward propagated through the hidden layer and transmitted to the output layer. If the output of the output layer is inconsistent with the expected output, the error is reversely propagated, and the error is used to adjust the weights of each unit in each layer. Forward and reverse propagation are alternately performed to continuously adjust the weights until the error between the output value and the expected value reaches the requirement or the set number of learning times is reached, and the network training is stopped.

[0079] The algorithm recognition module 13 realizes target recognition based on the principle of reverse neural network. It has fast information processing speed and large information storage capacity, and allows the mapping relationship between input and output to be obtained through self-learning and self-adaptation when the theory is insufficient, and finally recognizes the shape of the characteristic target through the orbital angular momentum spiral spectrum of the light field.

[0080] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that the features described in conjunction with a single embodiment may be used in other described embodiments.

Claims

1. A target feature recognition method based on a back propagation neural network and orbital angular momentum characteristic spectrum, characterized in that: The method comprises the following steps: Step 1: The vortex beam is irradiated onto a typical object in an atmospheric turbulence environment through a transmitting system to generate an echo signal; Step 2: performing phase distortion correction on the echo signal to obtain a correction light beam; Step 3, obtaining the target angular spectrum of the correction beam; Step 4: Input the target angular spectrum into the back propagation neural network to identify the target shape of a typical object.

2. The target feature recognition method based on back propagation neural network and orbital angular momentum characteristic spectrum according to claim 1, characterized in that: The phase distortion correction process in step 2 is: Step 21: Using a wavefront sensor to obtain the distorted light beam light field; Step 22: input the distorted light beam field and the original light beam field into the feedback signal controller together, and use the GS algorithm to obtain the original light beam phase in the feedback signal control; Step 23: Use the reverse phase of the original light beam to control the phase loading of the spatial light modulator, thereby correcting the distorted light beam and outputting the corrected light beam.

3. The target feature recognition method based on back propagation neural network and orbital angular momentum characteristic spectrum according to claim 2, characterized in that: The process of obtaining the target angular spectrum of the correction beam in step 3 is: The spatial light modulator is used to demodulate the correction beam light field, and then the optical power detector is used to detect the power of each orbital angular momentum mode, and finally the orbital angular momentum spiral spectrum of the correction beam light field is obtained as the target angular spectrum.

4. The target feature recognition method based on back propagation neural network and orbital angular momentum characteristic spectrum according to claim 3 is characterized in that: In step 4, the training back propagation neural network contains an input layer, a hidden layer, and an output layer. The input signal of the back propagation neural network is the target angular spectrum of the echo signal of a typical object in an atmospheric turbulence environment, and the output signal is the shape of the typical object. The training process is: Step 4: Determine the number of layers of the network, the number of neurons in each hidden layer, and randomly assign values ​​to the weights and thresholds of the network; Step 42: Input a training sample, select an activation function, calculate the network, and obtain the output value of the network; Step 43: Calculate the error between the output value and the expected value; if the error meets the requirement, stop training the network; if the error does not meet the requirement, execute step 44; Step 44: Back propagation, adjust the weights and thresholds of the network; Then, input the next training sample and return to step 42 for loop calculation.

5. The target feature recognition method based on back propagation neural network and orbital angular momentum characteristic spectrum according to claim 1, characterized in that: The atmospheric turbulence environment is simulated by the phase screen method. The atmospheric turbulence process is replaced by a random phase screen set with equal spacing Δz and simulated by the power spectrum inversion method. The light beam is freely transmitted between the random phase screens according to the angular spectrum diffraction theory. Using the modified VonKarman spectrum, the phase screen can be expressed as: Where (x, y) is the coordinate of the phase screen xoy plane, x = jΔx, y = lΔy, Δx, Δy are the sampling intervals of the x-axis and y-axis, j, l are integers, and h(κ x ,κ y ) is a Gaussian random number with unit variance and zero mean; Φ φ (f) is the phase power spectrum density of the phase screen, which can be obtained by the following formula: Where, f is the spatial frequency, f = κ / 2π, κ represents the three-dimensional space wave number; L0 is the outer scale of turbulence; l0 is the inner scale of turbulence; r0 is the atmospheric coherence length, is the turbulence intensity, z is the z-axis coordinate of the phase screen rectangular coordinate system, and Δz is the z-axis sampling interval.

6. A target feature recognition device based on a back propagation neural network and an orbital angular momentum characteristic spectrum, the device being used to implement the method described in any one of claims 1 to 5, characterized in that: The device comprises a vortex beam generating unit, a 4F system (4), a transmitting optical system (5), a No. 1 atmospheric turbulence simulator (6), a No. 1 beam splitter (7), a No. 2 atmospheric turbulence simulator (9), a receiving system (10), a distortion correction module (11), an angular spectrum analysis module (12) and an algorithm recognition module (13); A single-order vortex beam is output by a vortex beam generating unit. A 4F system (4) selects the first diffraction order of the single-order vortex beam and transmits it through a transmitting optical system (5). The beam is split through a No. 1 atmospheric turbulence simulator (6) and then through a No. 1 beam splitter (7) to irradiate a typical object (8) to generate an echo signal. A receiving system (10) receives the echo signal after passing through a No. 2 atmospheric turbulence simulator (9). A distortion correction module (11) performs phase correction on the distorted beam in the echo signal. An angular spectrum analysis module (12) obtains a target angular spectrum. An algorithm recognition module (13) recognizes the target shape.

7. The target feature recognition device based on back propagation neural network and orbital angular momentum characteristic spectrum according to claim 6, characterized in that: The vortex beam generating unit comprises a laser (1), a single-mode optical fiber (2) and a first spatial light modulator (3). The laser (1) emits laser light to the single-mode optical fiber (2), the single-mode optical fiber (2) collimates the light beam, and then the first spatial light modulator (3) performs orbital angular momentum quantum control on the light beam to generate a single-order vortex beam signal.

8. The target feature recognition device based on back propagation neural network and orbital angular momentum characteristic spectrum according to claim 6, characterized in that: The distortion correction module (11) comprises a second spatial light modulator (11-1), a second beam splitter (11-2), a wavefront sensor (11-3) and a feedback signal controller (11-4); the distorted light beam is split into two beams by the second beam splitter (11-2), one of which is detected by the wavefront sensor (11-3) to output a distorted light beam light field, which is input into the feedback signal controller (11-4) together with the original light beam light field; the feedback signal controller (11-4) outputs the original light beam phase, and loads the reverse phase of the original light beam into the second spatial light modulator (11-1); the other beam output by the second beam splitter (11-2) is subjected to phase distortion correction by the second spatial light modulator (11-1) to output a corrected light beam.

9. The target feature recognition device based on back propagation neural network and orbital angular momentum characteristic spectrum according to claim 6, characterized in that: The angular spectrum analysis module (12) comprises a No. 3 spatial light modulator and an optical power detector. The correction light beam is input into the No. 3 spatial light modulator to demodulate the light field. The demodulated light beam is input into the optical power detector. The orbital angular momentum spiral spectrum of the light field is obtained as the target angular spectrum according to the power of each orbital angular momentum mode detected by the optical power detector.

10. The target feature recognition device based on back propagation neural network and orbital angular momentum characteristic spectrum according to claim 9, characterized in that: The algorithm recognition module (13) is implemented by a back-propagation neural network, and recognizes the characteristic shapes of typical objects through the orbital angular momentum spiral spectrum of the light field. The characteristic shapes of the typical objects include triangles, squares and regular pentagons.

Citation Information

Cited By

  • Multilayer phase screen synthesis method for atmosphere turbulence simulation

    CN120469070A

  • A multi-layer phase screen synthesis method for atmospheric turbulence simulation

    CN120469070B

  • Optical fiber identification information determination method and device, electronic equipment and storage medium

    CN121567203A