Focusing regulation and control method and device based on vortex structure light field penetrating large-scale scattering medium

Abstract

The invention relates to a focusing regulation and control method and device based on a vortex structure light field penetrating through a large-scale scattering medium, and the method specifically comprises the steps: generating a collimated and expanded laser beam through a laser, carrying out the phase modulation of the laser beam through a spatial light modulator, and collecting a transmission speckle image related to the scattering medium through the combination of an sCMOS camera; and obtaining an optimal phase according to a focusing index quantity fed back by a focusing area by combining a genetic neural network algorithm with the vortex structure light field, and realizing speckle focusing through multiple iterations. Compared with the prior art, a vortex structure light field method is introduced to improve the feedback focusing regulation and control effect, the problem that a traditional feedback modulation algorithm cannot focus a large-scale scattering medium is solved, and the focusing intensity is remarkably superior to that of a lens focusing method and a wavefront shaping focusing method based on Gaussian beams.

Description

Technical Field

[0001] The present invention relates to the field of wavefront shaping laser control technology, and in particular to a focusing control method and device based on a vortex structure light field passing through a large-scale scattering medium. Background Art

[0002] Scattering phenomena often occur in scenes such as heavy fog, rain, snow, turbid water and smoke, which has a serious impact on traffic safety, underwater rescue, fire rescue and satellite remote sensing. Due to the uneven distribution of refractive index inside the scattering medium, the light wave will undergo strong multiple scattering during internal transmission, resulting in blurred images and complete loss of target information. At the same time, it also means that the current imaging method through static scattering media cannot solve the problem of macroscopic large-scale scattering medium imaging, which has a significant negative impact on traffic safety, underwater detection and biomedical imaging.

[0003] The purpose of this technology is to achieve the focusing of light beams through scattering media such as deep biological tissues, turbid liquids, and atmospheric turbulence. Existing iterative optimization techniques and their variants are only used to focus light through thin scattering media such as white paint layers, eggshells, frosted glass, and biological tissues. The applicable scenarios are very limited and cannot effectively improve the focusing and imaging problems in extreme rainy and foggy weather. The invention patent with application number CN201910563496.6 is based on the phase recovery algorithm and applied to small-scale dynamic scattering media such as rotating frosted glass, skim milk, a mixed solution of partially skimmed milk and water, and zinc oxide solution for imaging. This method is more effective for processing small-scale scattering media, but it is unable to reconstruct the target information through large-scale scattering media; the invention patent with application number CN202210891701.3 is based on the emerging deep learning method and applied to fat emulsion solution, artificial fog, and milk solution for imaging. This method shows good descattering ability, but it is limited to processing in mist scenes, and the processing process is more complicated; the invention patent with application number CN 202310711350.8 is based on the speckle autocorrelation method and applied to rotating frosted glass for focusing and imaging. Due to the long iteration time, its versatility is still insufficient, especially in scenes under extreme rain and fog conditions, where the effective feature information of the target is very limited. Therefore, the development of focusing control methods suitable for macroscopic large-scale scattering media still faces huge challenges. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method and device for focusing control of a vortex structure light field through a large-scale scattering medium, by which the wavefront phase of the laser input in front of the scattering medium is controlled by a spatial light modulator, so that the speckle image formed through the scattering medium changes with the change of the wavefront phase.

[0005] To achieve the above purpose, the present invention adopts the following technical solution: a method for focusing and controlling a vortex structure light field through a large-scale scattering medium, comprising the following steps:

[0006] Step 1, generating a vortex structure light field; loading the vortex phase into a spatial light modulator and combining it with circularly polarized light to generate a vortex structure light field;

[0007] Step 2, collecting a speckle image data set; the sCMOS camera collects speckle images passing through the scattering medium and builds an image data set;

[0008] Step 3: training a single-layer neural network; training the mapping relationship between the phase mask and the speckle image as a training model;

[0009] Step 4: Pre-focusing of the single-layer neural network; loading the phase mask obtained by the training model into the spatial light modulator to test the effectiveness of the training model;

[0010] Step 5, iterative optimization of genetic neural network in vortex structure light field: the predicted phase mask is combined with genetic algorithm to achieve speckle focusing through multiple iterations in the vortex structure light field.

[0011] In a preferred embodiment, the spatial light modulator is a reflective phase spatial light modulator.

[0012] In a preferred embodiment, the phase modulation of the spatial light modulator is expressed as:

[0013]

[0014] Among them, x n is the x-direction position of the n-th pixel area of ​​the spatial light modulator, y n is the y-direction position of the nth pixel area of ​​the spatial light modulator, E in (x n ,y n ) is the light field input to the spatial light modulator, φ n The amount of phase delay added to the nth pixel region of the spatial light modulator, E out (x n ,y n ) is the modulated light field, and i is an imaginary unit.

[0015] In a preferred embodiment, the phase of the vortex beam is a spiral phase, and the electric field expression of the vortex beam is:

[0016] E=E0exp(ilθ)exp(-ikz)

[0017] Among them, E0 is the amplitude, θ is the azimuth, k is the wave number, exp(ilθ) is the phase factor, l is the topological charge, and z is the propagation distance.

[0018] In a preferred embodiment, the water mist concentration is characterized by an optical thickness range, and the relationship expression is:

[0019] I=I0·e (-μL) =I0·e (-τ)

[0020] Among them, I0 is the initial incident light intensity, L is the thickness of the scattering medium, τ is the optical thickness, and μ is the extinction coefficient.

[0021] In a preferred embodiment, the data set in step 2 includes a mapping relationship between a speckle image and a phase mask.

[0022] In a preferred embodiment, the light field expression of the mapping relationship is:

[0023] X=F -1 (Y)

[0024] Among them, Y is the light field of the scattered light spot, and X is the light field in front of the scattering medium.

[0025] In a preferred embodiment, in step 4, a predicted phase mask is output based on the trained model and loaded into the spatial light modulator to achieve pre-focusing.

[0026] In a preferred embodiment, the genetic neural network in step 5 is a hybrid algorithm of a single-layer neural network and a genetic algorithm; the optimal phase mask predicted by the single-layer neural network in step 5 is iteratively optimized by the genetic algorithm.

[0027] The present invention also provides a focusing device based on a vortex structure light field passing through a large-scale scattering medium, and operates a focusing control method based on a vortex structure light field passing through a large-scale scattering medium; comprising: a laser, an attenuation plate, an aperture, a beam expander, an aperture, a lens, a spatial filter, an aperture, a reflective spatial light modulator, a reflector, a polarizer, a reflective spatial light modulator, a quarter-wave plate, a lens, a lens, a scattering medium and a sCMOS camera arranged in sequence along the light path direction;

[0028] The laser, attenuator, aperture, beam expander, aperture, lens, and spatial filter are used to generate a coherent light beam;

[0029] The aperture and reflective spatial light modulator are used to control the phase of the coherent light beam;

[0030] The polarizer, reflective spatial light modulator, and quarter wave plate are used to generate a circularly polarized vortex light beam;

[0031] The lens is used to compress the light spot of the light beam;

[0032] The scattering medium and the sCMOS camera are used to image the speckle to form a speckle image.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] 1) The present invention utilizes the correlation between the speckle image and the focusing control wavefront phase, uses the current speckle image to predict the wavefront phase required for the current speckle focusing, combines the excellent characteristics of the vortex structure light field passing through the large-scale scattering medium, completes the learning process of the genetic neural network through multiple iterations and obtains the optimal focusing wavefront phase of the speckle image.

[0035] Traditional focusing technology through scattering media is still limited to thin scattering media, and it is basically difficult to achieve focusing for large-scale scattering media in real scenes (clouds, turbid water, dense fog, etc.). The above solution, based on the vortex structure light field combined with machine learning to search the wavefront phase, solves the problem of difficulty in focusing in the task of speckle control through large-scale scattering media.

[0036] The existing feedback iteration methods in this field converge quickly but are prone to fall into local optimality. Based on this, the learnability of machine learning can provide a pre-focused wavefront phase in this control task, which facilitates the genetic algorithm to achieve rapid convergence and is not easily trapped in local optimality. At the same time, this method can work normally under normal parameters.

[0037] 2) The present invention uses different speckle images and corresponding learned wavefront phases to update the method. When light source fluctuations and other external environmental interferences occur in the acquisition light path, causing the illumination light field to change, the method can still have a good focusing effect on the existing speckle image and has good anti-interference ability.

[0038] 3) The scattering medium focusing control device based on reinforcement learning of the present invention has a simple structure, high stability and low cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Schematic diagram of the structure of a focusing control device based on a vortex structure light field through a large-scale scattering medium provided in an embodiment of the present invention.

[0040] Figure 2 is the kinoform diagram with a topological charge of 1 loaded on the reflective spatial light modulator.

[0041] Figure 3 (a) is the speckle pattern corresponding to the target, (b) is the focusing control diagram of the Gaussian beam (topological charge is 0), and (c) is the focusing control diagram of the vortex beam (topological charge is 1).

[0042] Figure 4 Comparison of the curves of the central section corresponding to the focus when the topological charge is 0 and 1. DETAILED DESCRIPTION

[0043] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0044] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0045] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application; as used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or their combinations.

[0046] refer to Figure 1-4 This embodiment provides a focusing control method based on a vortex structure light field passing through a large-scale scattering medium, including:

[0047] Step 1: Generation of vortex structure light field: Load the required vortex phase into the spatial light modulator and combine it with circular polarization to form a vortex structure light field;

[0048] like Figure 2 As shown, the vortex phase is a spiral phase diagram with a topological charge number of 1; the change of the polarization state of the vortex light beam, that is, rotating the quarter wave plate 13, is to change its angle to obtain a light beam with a desired polarization state distribution.

[0049] Step 2: Collect speckle image data set: The sCMOS camera collects speckle images through the scattering medium and builds its own image data set;

[0050] It should be noted that the spiral phase is not loaded in the acquisition process of the speckle image data set.

[0051] Step 3, training of a single-layer neural network: training the mapping relationship between the phase mask and the speckle image as a training model;

[0052] It should be noted that the training of the single-layer neural network aims to learn the mapping function between the speckle pattern and the phase mask, rather than directly modeling the scattering medium explicitly. Therefore, no test set was established for testing.

[0053] Step 4: Iterate and learn each speckle image based on the feedback control method based on machine learning;

[0054] The control method described in step 2 includes: a machine learning algorithm and a phase link for controlling the loading of the spatial light modulator.

[0055] Specifically, the feedback control method based on machine learning obtains real-time speckle images through speckle image acquisition, uses the current speckle image to search the focused wavefront phase of the machine learning algorithm, loads the searched focused wavefront phase by controlling the spatial light modulator, and finally combines the focusing condition of the speckle image of the vortex structure light field to calculate the reward value required for updating the reinforcement learning algorithm according to the focusing condition.

[0056] It should be noted that the spiral phase is not loaded in the speckle image acquisition link.

[0057] Each iteration outputs the optimal wavefront phase for focusing control searched by the current method;

[0058] The feedback control method based on reinforcement learning outputs the focusing wavefront phase required for focusing the current speckle image in each iteration.

[0059] Step 5: Use the spatial light modulator to load the searched wavefront phase and update the speckle image collected by the sCMOS camera. After repeated iterations, the optimal wavefront phase area searched by the method is stable and speckle focusing is achieved;

[0061] In an optional embodiment, a processor is used to transmit the focusing wavefront phase searched by a feedback control method based on reinforcement learning to a spatial light modulator, and a speckle image after loading the new wavefront phase is collected in real time by a sCMOS camera. The feedback control method stores historical records during the iteration process and uses the historical records to update the method, ultimately achieving the optimal wavefront phase required for real-time speckle focusing by inputting a real-time speckle image.

[0062] Specifically, the reward value that needs to be calculated in the feedback control method based on machine learning is the background signal-to-noise ratio, which is expressed as:

[0063]

[0064] Among them, f is the reward value, I n is the light intensity value of the pixel point in the target area, I m is the light intensity value of the pixel point in the background area, n is the total number of pixels in the target area, and m is the total number of pixels in the background area.

[0065] It should be noted that the definitions of the target area and the background area are:

[0066]

[0067] Among them, (x0, y0) is the center coordinate of the target point, r1 is the radius of the focused target area, and r2 is the radius of the focused background area.

[0068] The focusing control method based on the vortex structure light field through the large-scale scattering medium of the embodiment of the present invention utilizes the correlation between the wavefront phase and the speckle image, and performs learning and phase search on different speckle images through a machine learning algorithm under the vortex structure light field. By continuously updating the algorithm, a control method and a focusing wavefront phase suitable for the focusing of the speckle image of the experimental optical path are obtained.

[0069] In the focusing control method based on the vortex structure light field through the large-scale scattering medium of the embodiment of the present invention, a machine learning algorithm is used to learn the relationship between the speckle image and the focusing wavefront phase under the vortex structure light field to realize the focusing control of the speckle. Compared with the existing feedback control method for searching the local optimal wavefront phase and focusing on the large-scale scattering medium, the local optimal problem of the wavefront phase search is greatly improved, so that the focusing effect through the real scene (thick fog, etc.) is more stable and better, and it can quickly adapt to the new speckle image when the light source of the experimental light path fluctuates slightly, and has better universality.

[0070] like Figure 1 As shown, an embodiment of the present invention provides a focusing control device based on a vortex structure light field through a large-scale scattering medium, comprising: a solid laser 1, an attenuation plate 2, an aperture 3, a spatial filter 4, a lens 5, an aperture 6, a beam expander 7, an aperture 8, a spatial light modulator 9, a reflector 10, a polarizer 11, a spatial light modulator 12, a quarter-wave plate 13, a lens 14, a lens 15, a scattering medium 16, a sCMOS camera 17 and a processor 18 arranged in sequence along the light path direction.

[0071] The solid laser 1, the attenuation plate 2, the spatial filter 4, the lens 5 and the beam expander 7 are used to generate a uniform coherent light beam.

[0072] In this embodiment, a solid laser 1 is used, and the laser wavelength is 635nm; an attenuation plate 2 is used to weaken the laser light intensity to prevent the light intensity passing through the scattering medium from being too bright; a spatial filter 4 and a lens 5 are used to eliminate the stray light around the laser; a beam expander 7 and an aperture 8 are used to obtain a coherent light beam that can cover the entire center screen of the spatial light modulator.

[0073] The reflective spatial light modulator 9 is used to perform phase control on the coherent light beam.

[0074] The polarizer 11, the spatial light modulator 12 and the quarter-wave plate 13 are used to generate a vortex light beam with a desired polarization state.

[0075] In this embodiment, a circularly polarized laser beam is obtained by using a polarizer 11 and a quarter wave plate 13, and a vortex beam is generated by loading a spiral phase by a spatial light modulator 12. The angle difference between the polarizer 11 and the quarter wave plate 13 is 45 degrees.

[0076] The lens 14, the lens 15, the scattering medium 16, and the sCMOS camera 17 are used to image the speckle to form a speckle image.

[0077] In this embodiment, the lens is a convex lens, the focal length of lens 14 is 150 mm, the focal length of lens 15 is 27.5 mm, and lens 14 and lens 15 can make the light spot passing through the scattering medium smaller. The sCMOS camera adjusts the imaging focal plane to 150 mm behind the lens by focusing in order to obtain a clearer speckle image.

[0078] The processor is used to implement the regulation method and control the spatial light modulator and sCMOS camera.

[0079] In this embodiment, the processor is connected to the spatial light modulator and the sCMOS camera at the same time, allowing the spatial light modulator to load the focused wavefront phase and the sCMOS camera to collect the corresponding speckle image based on the focusing control method of the vortex structure light field through the large-scale scattering medium.

[0080] The focusing control device based on the vortex structure light field through the large-scale scattering medium in the embodiment of the present invention has a simple structure, high stability and low cost.

[0081] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A method for focusing and controlling a vortex structure light field through a large-scale scattering medium, characterized in that: The following steps are involved: Step 1, generating a vortex structure light field; loading the vortex phase into a spatial light modulator and combining it with circularly polarized light to generate a vortex structure light field; Step 2, collecting a speckle image data set; the sCMOS camera collects speckle images passing through the scattering medium and builds an image data set; Step 3: Training of single-layer neural network; Training the mapping relationship between the phase mask and the speckle image as a training model; Step 4: Pre-focusing of single-layer neural network; Load the phase mask obtained from the training model into the spatial light modulator to test the effectiveness of the training model; Step 5, iterative optimization of genetic neural network in vortex structure light field: the predicted phase mask is combined with genetic algorithm to achieve speckle focusing through multiple iterations in the vortex structure light field.

2. According to claim 1, a focusing control method based on a vortex structure light field passing through a large-scale scattering medium is characterized in that: The spatial light modulator is a reflective phase spatial light modulator.

3. According to claim 2, a method for focusing control of a vortex structure light field through a large-scale scattering medium, characterized in that: The phase modulation of the spatial light modulator is expressed as: Among them, x n is the x-direction position of the n-th pixel area of ​​the spatial light modulator, y n is the y-direction position of the nth pixel area of ​​the spatial light modulator, E in (x n ,y n ) is the light field input to the spatial light modulator, φ n The amount of phase delay added to the nth pixel region of the spatial light modulator, E out (x n ,y n ) is the modulated light field, and i is an imaginary unit.

4. The method for focusing and controlling a vortex structure light field through a large-scale scattering medium according to claim 1, characterized in that: The phase of the vortex beam is a spiral phase, and the electric field expression of the vortex beam is: E=E0exp(ilθ)exp(-ikz) Among them, E0 is the amplitude, θ is the azimuth, k is the wave number, exp(ilθ) is the phase factor, l is the topological charge, and z is the propagation distance.

5. The method for focusing and controlling a vortex structure light field through a large-scale scattering medium according to claim 1, characterized in that: The water mist concentration is characterized by the optical thickness range, and the relationship expression is: I=I0·e (-μL) =I0·e (-τ) Among them, I0 is the initial incident light intensity, L is the thickness of the scattering medium, τ is the optical thickness, and μ is the extinction coefficient.

6. The focusing control method based on vortex structure light field through large-scale scattering medium according to claim 1 is characterized in that: The data set in step 2 includes a mapping relationship between the speckle image and the phase mask.

7. The method for focusing and controlling a vortex structure light field through a large-scale scattering medium according to claim 6, characterized in that: The light field expression of the mapping relationship is: X=F -1 (Y) Among them, Y is the light field of the scattered light spot, and X is the light field in front of the scattering medium.

8. The method for focusing and controlling a vortex structure light field through a large-scale scattering medium according to claim 1, characterized in that: In step 4, a predicted phase mask is output based on the trained model and loaded into the spatial light modulator to achieve pre-focusing.

9. The method for focusing and controlling a vortex structure light field through a large-scale scattering medium according to claim 1, characterized in that: The genetic neural network in step 5 is a hybrid algorithm of a single-layer neural network and a genetic algorithm; the optimal phase mask predicted by the single-layer neural network in step 5 is iteratively optimized by the genetic algorithm.

10. A focusing device based on a vortex structure light field through a large-scale scattering medium, characterized in that: A method for focusing and controlling a vortex structure light field through a large-scale scattering medium according to any one of claims 1 to 9 above; comprising: a laser, an attenuator, an aperture, a beam expander, an aperture, a lens, a spatial filter, an aperture, a reflective spatial light modulator, a reflector, a polarizer, a reflective spatial light modulator, a quarter wave plate, a lens, a lens, a scattering medium and a sCMOS camera arranged in sequence along the light path direction; The laser, attenuator, aperture, beam expander, aperture, lens, and spatial filter are used to generate a coherent light beam; The aperture and reflective spatial light modulator are used to control the phase of the coherent light beam; The polarizer, reflective spatial light modulator, and quarter wave plate are used to generate a circularly polarized vortex light beam; The lens is used to compress the light spot of the light beam; The scattering medium and the sCMOS camera are used to image the speckle to form a speckle image.

Images