Focusing regulation and control device for transmitting large-scale scattering water body based on Airy structure light field and working method of focusing regulation and control device

Abstract

The invention provides a focusing regulation and control device based on an Airy structure light field penetrating through a large-scale scattering water body and a working method thereof, and the device comprises an Airy structure light beam modulation module which comprises a laser, an attenuation sheet, a spatial light modulator, a beam expander, a diaphragm I, a reflective spatial light modulator SLMA, a reflector and a reflective spatial light modulator SLMB; wherein the mask improvement generation module comprises a diaphragm II, a lens Lens A, a lens Lens B, a lens Lens C, a scattering medium and a CCD (Charge Coupled Device) camera; wherein the intelligent algorithm iteration module adopts a hybrid genetic neural network to realize wavefront modulation and focusing penetrating through a scattering medium; the hybrid genetic neural network comprises a hybrid intelligent algorithm of a single-layer neural network and a genetic algorithm.

Description

Technical Field

[0001] This invention proposes a focusing and control device for large-scale scattering water bodies based on Airy structured light fields and its working method, which relates to the field of optical microscope imaging and optical manipulation technology. Background Technology

[0002] To address the challenges posed by multiple light scattering, current mainstream techniques fall into two main categories, aiming to focus scattered light onto any target location behind a thick sample. The first category relies on pre-compensated wavefront shaping, which pre-corrects the phase or intensity of the incident wave to compensate for distortions that occur as it passes through the medium. The second category is based on the principle of phase conjugation, generating a wavefront conjugate to the scattered wave and allowing it to propagate back through the medium, thus canceling the scattering effect and ultimately refocusing the energy at the target point. However, most of these techniques are currently primarily applicable to static scattering media. This is because, during focusing, the propagation behavior of light in the medium must be considered time-invariant, meaning the medium's properties must remain stable during measurement and compensation. For example, in pre-compensated wavefront shaping, time-consuming iterative optimization or transfer matrix measurement often results in the total computation time exceeding the decorrelation time of the speckle pattern, making it impossible to adapt to changes in the medium in real time. Furthermore, the response speed limitations of key hardware such as cameras and spatial light modulators further restrict the ability of these methods to achieve effective focusing in dynamic scattering media. Nevertheless, algorithmic research to improve focusing performance continues to advance. For example, hybrid strategies combining intelligent optimization algorithms with machine learning have shown significant potential. Specifically, combining genetic algorithms with deep convolutional neural networks, or connecting particle swarm optimization with a single-layer neural network, has been shown to outperform any single algorithm. These hybrid methods effectively overcome their respective limitations by complementing each other's strengths; however, a large number of high-order speckle patterns and uneven focal energy distribution still exist after focusing.

[0003] Currently, this adaptive optics method is inefficient in highly turbid media, where the light distribution is severely disturbed during propagation, resulting in speckle patterns on the imaging plane. Scientists have made considerable efforts to obtain a perfect focal point or extract effective information from highly inhomogeneous media, and have actually succeeded in imaging through scattering media. However, as the media thickness continues to increase, the speckle decorrelation time is much shorter than the adjustment speed of wavefront shaping techniques, and scattering intensifies, causing almost all existing techniques to only achieve focusing and imaging through thin, dynamically scattering media. For a long time, scientists have believed that scattering constitutes a fundamental limitation on the penetration depth and resolution of optical systems. Ballistic light refers to the portion of a beam that penetrates directly without colliding with media particles. However, the intensity of ballistic light decays exponentially with penetration depth, and in strong scattering environments (such as in skin tissue), the mean free path of photons is often only about 1 mm. Whether serpentine or ballistic photons undergo seemingly complex and random scattering, they still retain a large amount of information. How to enhance and extract this effective information remains a challenge for current wavefront shaping techniques. Summary of the Invention

[0004] In view of this, to fill the gaps and deficiencies in existing technologies, this invention proposes a focusing and control device for large-scale scattering water bodies based on Airy structured light fields and its operating method. This invention fully utilizes the unique transmission characteristics of Airy structured beams to enhance their backscattered light while significantly reducing random modulation interference in the target area through an innovative mask generation method. A superior hybrid intelligent algorithm is used to achieve high-quality focusing dominated by freely propagating ballistic light, quasi-ballistic light, and higher-order diffracted ballistic light.

[0005] This invention proposes a focusing and control device for large-scale scattering water bodies based on Airy structure light fields, characterized by comprising an Airy structure beam modulation module, a mask improvement generation module, and an intelligent algorithm iteration module; The Airy structure beam modulation module includes a laser, an attenuator, a spatial light modulator, a beam expander, an aperture, a reflective spatial light modulator (SLMA), a mirror, and a reflective spatial light modulator (SLMB). The mask improvement generation module includes an aperture two, a lens A, a lens B, a lens C, a scattering medium, and a CCD camera. The intelligent algorithm iteration module includes a hybrid genetic neural network for wavefront modulation and focusing through a scattering medium; the hybrid genetic neural network includes a hybrid intelligent algorithm combining a single-layer neural network and a genetic algorithm.

[0006] Furthermore, the Airy structure beam modulation module uses a reflective spatial light modulator (SLMA) to perform wavefront shaping on the original beam emitted by the laser and obtain a shaped beam. The Airy structure beam modulation module then uses a reflective spatial light modulator (SLMB) to superimpose the corresponding cubic phase pattern onto the shaped beam to generate an Airy structure beam.

[0007] Furthermore, the reflective spatial light modulator (SLMA) is a liquid crystal reflective spatial light modulator; when the reflective spatial light modulator (SLMA) is working, the phase modulation range of the regional pixels within its working range is [0, 2π).

[0008] Furthermore, the mask improvement generation module uses a beam with a centrally hollowed-out annular wavefront shape to achieve focusing through the scattering medium, thereby reducing background noise and optimizing the intensity uniformity of the focal point.

[0009] Furthermore, the optimal focal lengths for Lens A, Lens B, and Lens C are 20cm, 10cm, and 20cm, respectively.

[0010] This invention also proposes a working method for a focusing and control device for large-scale scattering water bodies based on Airy structured light fields, applied to a focusing and control device for large-scale scattering water bodies based on Airy structured light fields as described in any one of these inventions, characterized by comprising the following: Step S1: Generate an Airy structured light field using an Airy structured beam modulation module; then adjust the spatial position of the Airy structured light field, including selecting the center of the outer ring region with a certain resolution as the preset focal point, so that the main lobe of the Airy structured beam coincides with the preset focal point; Step S2: Using the mask improvement generation module, set a ring-shaped mask according to the intensity distribution of the cross-section of the initially generated Airy light field; Step S3: The intelligent algorithm iterative module is used to establish the dataset, including using a CCD camera to collect scattering images to build a self-built dataset, training a single-layer neural network based on the mapping relationship between the dataset and the corresponding phase mask; using several focusing patterns with increasing focal size gradients as templates to load into the single-layer neural network to generate corresponding phase masks; then, evaluating the best pre-focusing phase template based on the ratio of the average light intensity of the target area to the average light intensity of the background area as the initial iteration template for the genetic algorithm; finally, the genetic algorithm is used to continue iterative optimization to reach the optimal value while maintaining the Airy light field.

[0011] Further, step S1 includes the following: The method of generating an Airy structured light field using an Airy structured beam modulation module includes generating an Airy structured beam using an Airy structured beam modulation module. This includes using a reflective spatial light modulator (SLMA) to perform wavefront shaping on the original beam emitted by the laser and obtaining a shaped beam. The Airy structured beam modulation module uses a reflective spatial light modulator (SLMB) to superimpose a corresponding cubic phase pattern on the shaped beam to generate an Airy structured beam. The spatial position of the Airy structured light field is adjusted to ensure that the main lobe of the Airy structured beam coincides with the preset focal point, including selecting the center of the outer ring region that satisfies the optimal resolution of 500*500 as the preset focal point.

[0012] Further, step S2 includes the following: The mask improvement generation module sets up a ring-shaped mask based on the intensity distribution of the initially generated Airy light field cross-section. This includes setting the inner circle of the mask to be square with an optimal resolution of 32*32, and setting the outer circle of the mask to ensure that the main lobe of the Airy structure beam coincides with the preset focus point. The optimal resolution of the outer circle of the mask is 500*500.

[0013] Further, step S3 includes the following: Step S31: The process of establishing the dataset using the intelligent algorithm iterative module includes: first, loading a random phase map into the spatial light modulator and acquiring the corresponding speckle image; then, using a single-layer neural network to learn the mapping relationship between the speckle map and the phase mask; Step S32: Based on the mapping relationship, generate ideal light field distribution images corresponding to different target focal sizes, generate a preliminary mask by neural network inverse reasoning, and reload it to the spatial light modulator; select the best-performing neural network prediction mask by evaluating the η value; where η is the ratio of the average light intensity of the target area to the average light intensity of the background area, and the specific expression is: ; Among them I n and I m represents the light intensity value of the pixels in the target area and the background area, respectively, and n and m represent the total number of pixels in the target area and the background area, respectively.

[0014] Further, step S3 includes the following: Step S33: Select two neural network prediction phase masks, ma and pa, and combine their pattern information by template cross-combination to generate offspring masks; the formula for synthesizing offspring masks is as follows: offspring=ma·T + pa (1-T); Where, offspring represents offspring, and T represents a randomly generated binary mask with gray levels of only 0 and 255. The process of generating offspring can effectively inherit the characteristics of superior parents. Step S34: In the process of generating new masks using a hybrid genetic network algorithm, a mutation operation is introduced. The mutation rate is typically designed to adaptively decrease with the iteration process, and the expression is: ; Where P m , initial and P m , final γ represents the initial and final mutation rates, respectively; g is the current iteration number; and γ is the decay number, used to control the rate of decrease in the mutation rate.

[0015] The present invention has the following advantages: This invention fully utilizes the unique transmission characteristics of Airy structure beams to enhance their backscattered light while significantly reducing random modulation interference within the target area through an innovative mask generation method. A superior hybrid intelligent algorithm is employed to achieve high-quality focusing dominated by freely propagating ballistic beams, quasi-ballistic beams, and higher-order diffracted ballistic beams. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the device of the present invention.

[0017] Figure 2 This is a schematic diagram of the process for generating Airy light according to the present invention.

[0018] Figure 3 This is a schematic diagram illustrating the modified mask method of the present invention and the focusing result using Gaussian light as an example.

[0019] Figure 4 This is a flowchart illustrating the specific steps of the intelligent algorithm of the present invention.

[0020] Figure 5 This is a schematic diagram showing the average light intensity, background noise, and η value of the region of interest after focusing Airy beams with different directional coefficients according to the present invention.

[0021] Figure 6 This is a schematic diagram of the roundness calculation result of the focal point after modulation and focusing, and the focal point boundary obtained based on the Euclidean distance between pixels, according to the present invention.

[0022] Figure 7 This diagram illustrates the propagation model of Gaussian and Airy light through a random field and the numerical simulation results of cross-sectional light intensity in this invention. Figure 1 .

[0023] Figure 8 This diagram illustrates the propagation model of Gaussian and Airy light through a random field and the numerical simulation results of cross-sectional light intensity in this invention. Figure 2 .

[0024] Figure 9 This diagram illustrates the propagation model of Gaussian and Airy light through a random field and the numerical simulation results of cross-sectional light intensity in this invention. Figure 3 .

[0025] Figure 10 This is a schematic diagram of the normalized correlation curve over time for a 5cm cuvette containing a 3% skim milk aqueous solution according to the present invention.

[0026] Figure 11 This is a flowchart of the steps of the present invention.

[0027] In the diagram: 1-Laser, 2-Attenuator, 3-Spatial light modulator, 4-Beam expander, 5-Aperture 1, 6-Reflective spatial light modulator SLMA, 7-Mirror, 8-Reflective spatial light modulator SLMB, 9-Aperture 2, 10-Lens A, 11-Lens B, 12-Lens C, 13-Scattering medium, 14-CCD camera. Detailed Implementation

[0028] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings.

[0029] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0030] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention; as used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise; furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0031] like Figures 1 to 11 As shown, this invention proposes a focusing and control device for large-scale scattering water bodies based on Airy structured light fields and its working method, characterized by including the following: This invention proposes a focusing and control device for large-scale scattering water bodies based on Airy structure light fields, characterized by comprising an Airy structure beam modulation module, a mask improvement generation module, and an intelligent algorithm iteration module; The Airy structure beam modulation module includes a laser 1, an attenuator 2, a spatial light modulator 3, a beam expander 4, an aperture 5, a reflective spatial light modulator SLMA 6, a reflector 7, and a reflective spatial light modulator SLMB 8. The mask improvement generation module includes aperture 9, lens A10, lens B11, lens 12, scattering medium 13, and CCD camera 14. The intelligent algorithm iteration module includes a hybrid genetic neural network for wavefront modulation and focusing through a scattering medium; the hybrid genetic neural network includes a hybrid intelligent algorithm combining a single-layer neural network and a genetic algorithm.

[0032] Furthermore, the Airy structure beam modulation module uses a reflective spatial light modulator SLMA6 to perform wavefront shaping on the original beam emitted by the laser to obtain a shaped beam, and the Airy structure beam modulation module uses a reflective spatial light modulator SLMB8 to superimpose the corresponding cubic phase pattern on the shaped beam to generate an Airy structure beam.

[0033] Furthermore, the reflective spatial light modulator SLMA6 is a liquid crystal reflective spatial light modulator; when the reflective spatial light modulator SLMA6 is working, the phase modulation range of the regional pixels within the working range is [0, 2π).

[0034] Furthermore, the mask improvement generation module uses a beam with a centrally hollowed-out annular wavefront shape to achieve focusing through the scattering medium, thereby reducing background noise and optimizing the intensity uniformity of the focal point.

[0035] Furthermore, the optimal focal lengths for Lens A10, Lens B11, and Lens C12 are 20cm, 10cm, and 20cm, respectively.

[0036] like Figure 11 As shown, the present invention also proposes a working method for a focusing and control device for large-scale scattering water bodies based on Airy structured light fields, applied to a focusing and control device for large-scale scattering water bodies based on Airy structured light fields as described in any one of the present invention, characterized by comprising the following: Step S1: Generate an Airy structured light field using an Airy structured beam modulation module; then adjust the spatial position of the Airy structured light field, including selecting the center of the outer ring region with a certain resolution as the preset focal point, so that the main lobe of the Airy structured beam coincides with the preset focal point; Step S2: Using the mask improvement generation module, set a ring-shaped mask according to the intensity distribution of the cross-section of the initially generated Airy light field; Step S3: The intelligent algorithm iterative module is used to establish the dataset, including using a CCD camera to collect scattering images to build a self-built dataset, training a single-layer neural network based on the mapping relationship between the dataset and the corresponding phase mask; using several focusing patterns with increasing focal size gradients as templates to load into the single-layer neural network to generate corresponding phase masks; then, evaluating the best pre-focusing phase template based on the ratio of the average light intensity of the target area to the average light intensity of the background area as the initial iteration template for the genetic algorithm; finally, the genetic algorithm is used to continue iterative optimization to reach the optimal value while maintaining the Airy light field.

[0037] Further, step S1 includes the following: The method of generating an Airy structured light field using an Airy structured beam modulation module includes generating an Airy structured beam using an Airy structured beam modulation module. This includes using a reflective spatial light modulator (SLMA) to perform wavefront shaping on the original beam emitted by the laser and obtaining a shaped beam. The Airy structured beam modulation module uses a reflective spatial light modulator (SLMB) to superimpose a corresponding cubic phase pattern on the shaped beam to generate an Airy structured beam. The spatial position of the Airy structured light field is adjusted to ensure that the main lobe of the Airy structured beam coincides with the preset focal point, including selecting the center of the outer ring region that satisfies the optimal resolution of 500*500 as the preset focal point.

[0038] Further, step S2 includes the following: The mask improvement generation module sets up a ring-shaped mask based on the intensity distribution of the initially generated Airy light field cross-section. This includes setting the inner circle of the mask to be square with an optimal resolution of 32*32, and setting the outer circle of the mask to ensure that the main lobe of the Airy structure beam coincides with the preset focus point. The optimal resolution of the outer circle of the mask is 500*500.

[0039] Further, step S3 includes the following: Step S31: The process of establishing the dataset using the intelligent algorithm iterative module includes: first, loading a random phase map into the spatial light modulator and acquiring the corresponding speckle image; then, using a single-layer neural network to learn the mapping relationship between the speckle map and the phase mask; Step S32: Based on the mapping relationship, generate ideal light field distribution images corresponding to different target focal sizes, generate a preliminary mask by neural network inverse reasoning, and reload it to the spatial light modulator; select the best-performing neural network prediction mask by evaluating the η value; where η is the ratio of the average light intensity of the target area to the average light intensity of the background area, and the specific expression is: ; Among them I n and I mrepresents the light intensity value of the pixels in the target area and the background area, respectively, and n and m represent the total number of pixels in the target area and the background area, respectively.

[0040] Further, step S3 includes the following: Step S33: Select two neural network prediction phase masks, ma and pa, and combine their pattern information by template cross-combination to generate offspring masks; the formula for synthesizing offspring masks is as follows: offspring=ma·T + pa (1-T); Where, offspring represents offspring, and T represents a randomly generated binary mask with gray levels of only 0 and 255. The process of generating offspring can effectively inherit the characteristics of superior parents. Step S34: In the process of generating new masks using a hybrid genetic network algorithm, a mutation operation is introduced. The mutation rate is typically designed to adaptively decrease with the iteration process, and the expression is: ; Where P m , initial and P m , final γ represents the initial and final mutation rates, respectively; g is the current iteration number; and γ is the decay number, used to control the rate of decrease in the mutation rate.

[0041] In addition to the above, the present invention also has related embodiments, including the following: In one embodiment of this invention, an Airy structured beam modulation module is proposed. Based on the high controllability of structured beams, the advantages of corresponding structured beams are matched to the focusing performance requirements of different application scenarios, which has significant research value. In comprehensive evaluation, the Airy beam exhibits the best focusing performance. In this module, this invention combines the advantages of the Airy beam with an intelligent iterative algorithm to achieve excellent focusing results.

[0042] In one embodiment of the present invention, an improved mask generation module is proposed: In a previous focusing experiment on turbid water based on an intelligent optimization algorithm, to reduce the optical path, the present invention improves the experimental optical path as follows: Figure 1As shown, after optical path adjustment, diffracted light with a hollow intensity distribution is observed (this structure originates from wavefront broadening and lateral energy diffusion caused by diffraction). This type of beam exhibits excellent focusing ability through the scattering medium, indicating that a specific wavefront shape significantly enhances the focusing effect. Using this annular wavefront shape with a central cavity for focusing through the scattering medium significantly reduces background noise and optimizes the intensity uniformity of the focal point. This invention defines the focusing focal region as the Region of Interest (ROI). Research suggests that the randomness of algorithm iteration during focusing can cause beams originally within the ROI to be modulated outside the region. This suggests that improvements in mask generation could avoid this algorithmic randomness.

[0043] In one embodiment of this invention, an intelligent algorithm iteration module is proposed: this module is the core part of wavefront shaping for focusing. This invention fully combines the advantages of two methods. To fully leverage the efficiency of neural networks in learning complex mapping relationships and the robustness of genetic algorithms in global search optimization, a new paradigm of phased, collaboratively optimized wavefront modulation is formed. This invention employs a hybrid intelligent algorithm combining a single-layer neural network (SLNN) and a genetic algorithm (GA)—a genetic neural network (GNN)—to achieve wavefront modulation and focusing through scattering media.

[0044] In one embodiment of the present invention, the workflow of the Airy structured beam modulation module includes the following: An Airy beam is a non-traditional optical field with a unique spatial structure and propagation characteristics. Its name derives from its mathematical form, which is based on the Airy function. Ideally (i.e., with infinite energy), the electric field distribution of a one-dimensional Airy beam can be expressed as: ; Here, Ai(·) represents the Airy function. Since the ideal Airy function has infinite energy, a perfectly ideal Airy beam cannot be realized in reality. To better describe a practically achievable Airy beam, a truncation factor 'a' is usually introduced to achieve "truncation." This study uses a spatial light modulator with a cubic phase pattern loaded onto it, followed by a Fourier transform lens to generate an Airy beam. Given that the beam's transverse intensity distribution and propagation characteristics have a crucial influence on the scattering process, the experiment uses the direction coefficient Y as a variable to systematically study its relationship with focusing quality. Figure 1 In the experimental optical path, this invention utilizes SLM1 to perform wavefront shaping on the beam, while the function of SLM2 is to superimpose the corresponding cubic phase pattern to generate structured light. Figure 2 The cubic phase pattern being displayed requires selecting the most suitable direction coefficient based on the size of the original light spot.

[0045] Airy beams, lacking orbital angular momentum (OAM), still exhibit unique dynamic behavior during propagation, giving them a potential advantage in resisting scattering effects. Firstly, they possess self-acceleration: as an Airy beam propagates in a certain direction, its main lobe undergoes a lateral shift in its parabolic trajectory with increasing propagation distance. Secondly, and importantly, Airy beams exhibit self-healing properties. This means that when the beam is partially obstructed or disturbed, it can recover its original intensity profile after continuing propagation for a certain distance. Structured light with self-healing properties exhibits excellent anti-scattering capabilities, but therefore, when using it for focusing, it is essential to ensure that the target focal point coincides with the main lobe position; otherwise, the main lobe will "reconstruct" due to the self-healing mechanism. Therefore, this invention selects the center of a 500×500 area as the focal point, i.e., coordinates (250, 250), and uses fine-tuning to ensure that the Airy beam's main lobe coincides with the center position; otherwise, the main lobe will self-heal, resulting in two focal points. Figure 2 .

[0046] In one embodiment of the present invention, the workflow of the mask improvement generation module includes the following: 1) Set up a focusing device for large-scale scattering water bodies based on Airy structured light fields, such as... Figure 1 It includes: a laser, a reflective spatial light modulator (SLMA), a mirror, a reflective spatial light modulator (SLMB), an aperture (S), a lens (LensA) with a focal length of 20cm, a lens (LensB) with a focal length of 10cm, a lens (LensC) with a focal length of 20cm, a scattering medium, and a CCD camera arranged sequentially along the optical path.

[0047] 2) The core wavefront shaping device used in this invention is a liquid crystal spatial light modulator SLM1 (LC-SLM). Although its native resolution is 1920×1080, in order to ensure modulation efficiency, a 1024×1024 pixel area in the center is actually selected for wavefront modulation. The phase modulation range of the pixels in this area is [0, 2π].

[0048] 3) To ensure the ROI region is unaffected by spatial light modulation, a 32×32 pixel unmodulated area (grayscale value of 0) is set at the center of the final phase mask generated by the hybrid intelligent algorithm. That is, instead of generating a 1024×1024 square phase mask, only a 1024×1024 square annular mask missing the center (32×32 pixels) needs to be generated. Figure 4 As shown, this improves focusing quality while reducing mask generation time.

[0049] 4) The optimal phase for a single channel is determined by searching within a cyclic phase range from 0 to 2π. For each channel, the phase value with the highest target intensity is recorded. Results using the improved mask generation method in experiments focusing Gaussian light through a scattering medium are shown below. Figure 3 As shown, the maximum light intensity of the focal point before changing the phase mask generation method was 26036, and the maximum light intensity of the focal point after the change was 29823. This processing increased the intensity of the focal point by 14.6%, while the intensity of the speckle in the vicinity decreased by 20%, effectively verifying the feasibility of this approach.

[0050] In one embodiment of the present invention, the workflow of the intelligent algorithm iteration module includes the following: The system principle block diagram of the genetic neural network used in the intelligent algorithm iteration module is as follows: Figure 4 As shown, this method innovatively combines the advantages of neural networks and genetic algorithms. The execution flow first loads a random phase map into a spatial light modulator (SLM) and acquires the corresponding speckle image; then, it uses a spatial neural network (SLNN) to learn the mapping relationship between the speckle map and the phase mask. In dynamic scattering media, the time-varying characteristics of the medium make its microstructure extremely sensitive to light. In this case, if extensive training is performed, the obtained mapping model is often merely a statistical average of the real physical model, resulting in significant bias. Therefore, for single-channel optimization in dynamic media, the conventional precise SLNN method is no longer applicable and needs to be improved into an approximate SLNN method capable of coarse focusing.

[0051] Based on the mapping relationship, ideal light field distribution images corresponding to different target focal sizes are generated. A preliminary mask is generated by neural network inverse reasoning and reloaded into the SLM. The best-performing neural network prediction mask is selected through η value evaluation. η is the ratio of the average light intensity of the target region to the average light intensity of the background region, specifically expressed as: ; Where In and Im represent the light intensity values ​​of pixels in the target area and background area, respectively, and n and m represent the total number of pixels in the target area and background area, respectively.

[0052] The target area and background area are defined as follows: ; ; Two neural network prediction phase masks (labeled ma and pa, respectively) are selected, and their pattern information is combined according to certain rules to generate offspring. A common breeding method is template crossover, which uses a breeding template T to synthesize offspring masks according to the following formula: offspring=ma·T + pa (1-T); This process effectively inherits the characteristics of superior parents. To further improve population diversity and convergence performance, the algorithm introduces a mutation operation into the newly generated mask, that is, randomly changing some of its phase elements. To avoid the mutation process destroying existing superior patterns, the mutation rate is usually designed to adaptively decrease with the iteration process, specifically expressed as: ; Where P m,infinal and P m,final γ represents the initial and final mutation rates, respectively; g is the current iteration number; and γ is the decay number, used to control the rate of decrease in mutation rate. The SLNN structure contains only one fully connected layer, featuring fast training speed and strong generalization ability. It effectively reduces the risk of error accumulation and getting trapped in local optima during iteration, achieving high-quality pre-focusing results in a short time and providing a reliable search starting point for subsequent genetic algorithms.

[0053] In one embodiment of the present invention, the specific implementation of the large-scale scattering water focusing strategy is as follows: The focusing process consists of two stages. The first stage is the generation of the Airy structured light field: the cubic phase distribution map of the two-dimensional Airy light field is loaded onto the spatial light modulator to modulate the coherent light field into an Airy structured light. Then, the spatial position of the Airy light field is adjusted to ensure that the main lobe of the Airy beam coincides with the preset focal point.

[0054] The second stage is the wavefront shaping stage: First, based on the intensity distribution of the initially generated Airy light field cross-section, a mask is set as a ring-shaped structure with an inner circle of 0×0 to 32×32 and an outer circle of 500×500. Then, the CCD camera 14 acquires scattering images to create a self-built dataset. A single-layer neural network is trained based on the mapping relationship between the dataset and the corresponding phase mask. Six focusing patterns with increasing focal size gradients are loaded into the single-layer neural network as templates to generate the corresponding phase mask. Then, the optimal pre-focused phase template is evaluated based on the η value and used as the initial iteration template for the genetic algorithm. The genetic algorithm continues to iterate and optimize under the Airy light field until the optimal result is achieved.

[0055] The specific experimental details are as follows: The single-layer neural network (SLNN) used in this invention learns the mapping relationship between the phase mask and the speckle pattern, rather than specifically modeling the scattering medium. Therefore, no test set was established for testing.

[0056] The iterative optimization of the genetic algorithm used in this invention is divided into two parts. The first part consists of one hundred iterations with the absolute value of the light intensity within the ROI region as the fitness function, and the second part consists of one hundred iterations with the ratio of the average light intensity of the ROI region to the average light intensity of the background region as the fitness function.

[0057] In addition to Airy beams, this invention also conducted focusing experiments on other types of structured beams. The focusing performance of two main categories of structured beams was quantitatively analyzed: one category consists of beams carrying orbital angular momentum (OAM), including Laguerre-Gaussian beams, Bessel beams, and perfect vortex beams; the other category consists of Airy beams without OAM. Airy beams modulated with appropriate spot size and direction factor showed the best performance, with η values ​​generally higher than 65, such as... Figure 5 As shown. This is mainly due to its inherent anti-interference capability; the speckle noise generated primarily originates from higher-order diffraction of the side lobes. The side lobes of an Airy beam propagate along a curved trajectory, allowing its energy flow to bypass obstructions in the transverse plane. This makes it suitable for scenarios with transverse interference such as pipe obstructions or local wavefront distortion caused by atmospheric turbulence. The self-healing property of an Airy beam stems from its propagation mechanism of continuously replenishing energy from the side lobes to the main lobe. This mechanism enables the beam to maintain its original intensity distribution well after passing through the scattering medium, and the energy flow direction remains relatively definite.

[0058] In terms of focal point morphology, self-healing structured light typically performs poorly. Due to the influence of the main lobe and side lobes, it is difficult to maintain an ideal circular and uniform distribution of the focused spot. Similarly, Bessel beams, which also possess self-healing properties, often exhibit a long strip distribution along the main lobe ring. In contrast, Airy beams have a main lobe morphology closer to a circle, resulting in superior roundness. The focusing results of Airy beams are shown below. Figure 6 As shown.

[0059] The Airy beam in this invention exhibits excellent anti-interference capabilities. This invention employs numerical simulation to study the propagation process of the structured beam in a scattering medium. By outputting the light intensity distribution at different locations, the propagation characteristics of the Airy beam under scattering conditions are analyzed. The results are as follows: Figure 7 , Figure 8 and Figure 9 As shown in the figure, this invention simulates the light intensity distribution of a Gaussian beam and an Airy beam propagating in a random scattering field at distances of L=50 cm, 100 cm, and 150 cm. The red line in the figure represents the normalized light intensity at the center line of the output image, while the black line represents the normalized light intensity distribution at the original center line of the input beam. The simulation results show that as the propagation distance in the random field increases, the light intensity of the Gaussian beam gradually decreases, and its distribution shape gradually deviates from the initial Gaussian profile. In contrast, the Airy beam exhibits excellent light intensity distribution maintenance and anti-interference capabilities: even with increased propagation depth, although the intensity of its side lobes weakens, the intensity of its main lobe remains stable, effectively reducing forward scattering.

[0060] The training time of the SLNN model in the genetic neural network algorithm used in this invention far exceeds the speckle decorrelation time of the medium used. Figure 10 As shown. In Figure 10 In the middle, g nThis represents the correlation coefficient between the speckle pattern at this moment and the speckle pattern in the first frame, and the speckle decorrelation time. Defined as a speckle correlation coefficient decreasing to 1 / e 2 (or 13.5%) the time required, g n < 1 / e 2 This indicates the decorrelation time of the scattering medium at this moment.

[0061] However, by combining genetic algorithms, pre-focusing of scattered light through turbid liquids was successfully achieved. Compared to existing mainstream methods for improving device performance by overcoming decorrelation of scattering media, the intelligent algorithm used in this invention is simpler and has a wider range of applications.

[0062] The above are preferred embodiments of the present invention. Any changes made to the technical solution of the present invention that do not exceed the scope of the technical solution of the present invention shall fall within the protection scope of the present invention.

Claims

1. A device for focusing and controlling large-scale scattering water bodies through Airy structured light fields, characterized in that, Includes an Airy structure beam modulation module, a mask improvement generation module, and an intelligent algorithm iteration module; The Airy structure beam modulation module includes a laser, an attenuator, a spatial light modulator, a beam expander, an aperture, a reflective spatial light modulator (SLMA), a mirror, and a reflective spatial light modulator (SLMB). The mask improvement generation module includes an aperture two, a lens A, a lens B, a lens C, a scattering medium, and a CCD camera. The intelligent algorithm iteration module includes a hybrid genetic neural network for wavefront modulation and focusing through a scattering medium; the hybrid genetic neural network includes a hybrid intelligent algorithm combining a single-layer neural network and a genetic algorithm.

2. The focusing and control device for large-scale scattering water bodies based on Airy structured light fields according to claim 1, characterized in that, The Airy structure beam modulation module uses a reflective spatial light modulator (SLMA) to perform wavefront shaping on the original beam emitted by the laser and obtain a shaped beam. The Airy structure beam modulation module uses a reflective spatial light modulator (SLMB) to superimpose the corresponding cubic phase pattern on the shaped beam to generate an Airy structure beam.

3. The focusing and control device for large-scale scattering water bodies based on Airy structured light fields according to claim 1, characterized in that, The aforementioned reflective spatial light modulator (SLMA) is a liquid crystal reflective spatial light modulator; when the SLMA is working, the phase modulation range of the regional pixels within its working range is [0, 2π].

4. The focusing and control device for large-scale scattering water bodies based on Airy structured light fields according to claim 1, characterized in that, The mask improvement generation module generates a beam with a circular wavefront shape having a central cavity to achieve focusing through the scattering medium, reduce background noise, and optimize the intensity uniformity of the focal point.

5. The focusing and control device for large-scale scattering water bodies based on Airy structured light fields according to claim 1, characterized in that, The optimal focal lengths for Lens A, Lens B, and Lens C are 20cm, 10cm, and 20cm, respectively.

6. A method for operating a focusing and control device for large-scale scattering water bodies based on Airy structured light fields, applied to the focusing and control device for large-scale scattering water bodies based on Airy structured light fields as described in any one of claims 1 to 5, characterized in that, Includes the following: Step S1: Generate an Airy structured light field using an Airy structured beam modulation module; then adjust the spatial position of the Airy structured light field, including selecting the center of the outer ring region with a certain resolution as the preset focal point, so that the main lobe of the Airy structured beam coincides with the preset focal point; Step S2: Using the mask improvement generation module, set a ring-shaped mask according to the intensity distribution of the cross-section of the initially generated Airy light field; Step S3: The intelligent algorithm iterative module is used to establish the dataset, including using a CCD camera to collect scattering images to build a self-built dataset, training a single-layer neural network based on the mapping relationship between the dataset and the corresponding phase mask; using several focusing patterns with increasing focal size gradients as templates to load into the single-layer neural network to generate corresponding phase masks; then, evaluating the best pre-focusing phase template based on the ratio of the average light intensity of the target area to the average light intensity of the background area as the initial iteration template for the genetic algorithm; finally, the genetic algorithm is used to continue iterative optimization to reach the optimal value while maintaining the Airy light field.

7. The working method of the focusing and control device for large-scale scattering water body based on Airy structured light field according to claim 6, characterized in that, Step S1 includes the following: The method of generating an Airy structured light field using an Airy structured beam modulation module includes generating an Airy structured beam using an Airy structured beam modulation module. This includes using a reflective spatial light modulator (SLMA) to perform wavefront shaping on the original beam emitted by the laser and obtaining a shaped beam. The Airy structured beam modulation module uses a reflective spatial light modulator (SLMB) to superimpose a corresponding cubic phase pattern on the shaped beam to generate an Airy structured beam. The spatial position of the Airy structured light field is adjusted to ensure that the main lobe of the Airy structured beam coincides with the preset focal point, including selecting the center of the outer ring region that satisfies the optimal resolution of 500*500 as the preset focal point.

8. The working method of the focusing and control device for large-scale scattering water body based on Airy structured light field according to claim 7, characterized in that, Step S2 includes the following: The mask improvement generation module sets up a ring-shaped mask based on the intensity distribution of the initially generated Airy light field cross-section. This includes setting the inner circle of the mask to be square with an optimal resolution of 32*32, and setting the outer circle of the mask to ensure that the main lobe of the Airy structure beam coincides with the preset focus point. The optimal resolution of the outer circle of the mask is 500*500.

9. The working method of the focusing and control device for large-scale scattering water body based on Airy structured light field according to claim 8, characterized in that, Step S3 includes the following: Step S31: The process of establishing the dataset using the intelligent algorithm iterative module includes: first, loading a random phase map into the spatial light modulator and acquiring the corresponding speckle image; then, using a single-layer neural network to learn the mapping relationship between the speckle map and the phase mask; Step S32: Based on the mapping relationship, generate ideal light field distribution images corresponding to different target focal sizes, generate a preliminary mask by neural network inverse reasoning, and reload it to the spatial light modulator; select the best-performing neural network prediction mask by evaluating the η value; where η is the ratio of the average light intensity of the target area to the average light intensity of the background area, and the specific expression is: ； Among them I n and I m represents the light intensity value of the pixels in the target area and the background area, respectively, and n and m represent the total number of pixels in the target area and the background area, respectively.

10. The working method of the focusing and control device for large-scale scattering water body based on Airy structured light field according to claim 9, characterized in that, Step S3 includes the following: Step S33: Select two neural network prediction phase masks, ma and pa, and combine their pattern information by template cross-combination to generate offspring masks; the formula for synthesizing offspring masks is as follows: offspring=ma·T + pa (1-T); Where, offspring represents offspring, and T represents a randomly generated binary mask with gray levels of only 0 and 255. The process of generating offspring can effectively inherit the characteristics of superior parents. Step S34: In the process of generating new masks using a hybrid genetic network algorithm, a mutation operation is introduced. The mutation rate is typically designed to adaptively decrease with the iteration process, and the expression is: ； Where P m , initial and P m , final γ represents the initial and final mutation rates, respectively; g is the current iteration number; and γ is the decay number, used to control the rate of decrease in the mutation rate.

Images