A high-dimensional OAM mode fiber sensing demodulation method based on buneman algorithm

Abstract

The application relates to the technical field of optical fiber sensing and optical field information processing, and discloses a high-dimensional OAM mode optical fiber sensing demodulation method based on a Buneman algorithm, which comprises the following steps: 1, acquiring an interference image; 2, pre-processing the image; 3, extracting bright spot extreme points; 4, carrying out sub-pixel level center positioning; performing local interpolation calculation on each point in the extracted initial bright spot center coordinate set to obtain a more accurate sub-pixel level center position; 5, performing polar coordinate conversion on the sub-pixel level center coordinates; converting the obtained sub-pixel level center coordinates of each bright spot into a polar coordinate form; 6, fitting and estimating a rotation angle; and 7, outputting an average rotation angle offset. The application is suitable for large dynamic range interferometry, has high demodulation accuracy, has sub-angle level demodulation capability, accuracy and linear stability under multi-angle conditions, and has a simple process and low calculation complexity.

Description

Technical Field

[0001] This invention relates to the field of fiber optic sensing and optical field information processing technology, and in particular to a high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm. Background Technology

[0002] Fiber optic sensing technology is widely used in current marine multi-parameter monitoring. However, when faced with the combined effects of multiple physical quantities, the sensing signals suffer from severe cross-sensitivity. Traditional solutions mainly rely on structural partitioning and material differences to achieve physical isolation, which limits the system's integration and stability.

[0003] OAM (orbital angular momentum) modes, as one of the intrinsic degrees of freedom of light, possess complex spatial structures and diverse modal responses, exhibiting a natural ability to carry multidimensional information. They have become a research hotspot for decoupling strategies in optical sensing in recent years. Interference images of OAM modes often exhibit "petal"-shaped fringe patterns. The rotation angle of these patterns is often related to external perturbations (such as temperature, salinity, and depth). Therefore, demodulating the angular changes of the petal fringes becomes an effective method for inverting physical parameters.

[0004] To extract rotation angle information, existing technologies mainly employ the following two approaches:

[0005] (1) Image polar coordinate transformation + one-dimensional angular intensity extraction + FFT frequency estimation:

[0006] After converting the interferometric image to polar coordinates, the angular intensity distribution sequence is extracted along a fixed radius. The sequence is then subjected to a Fast Fourier Transform (FFT), and the rotation angle is estimated by the position of the main frequency peak.

[0007] This method has a simple structure and is suitable for ideal symmetrical interferograms, but its accuracy is limited when there is noise, distortion, or non-ideal structure.

[0008] (2) Three-point interpolation methods (such as Buneman Frequency Estimation):

[0009] The main peak and its left and right neighboring points are extracted from the FFT result, and the sub-pixel position of the main frequency is estimated using a quadratic interpolation formula to improve demodulation accuracy.

[0010] However, this method is only applicable to one-dimensional sequences and is not robust to phenomena such as irregularity, distortion, and intensity asymmetry of interference fringes.

[0011] In addition, some studies have attempted to identify interference structures through image processing methods such as image center centroid tracking, template matching, and Hough transform. However, these methods mostly rely on manual parameter settings or specific structural templates, and lack versatility and accuracy, making it difficult to extend them to high-dimensional and complex scenarios.

[0012] While the above methods can achieve basic estimation of interferogram angles in some scenarios, they still have the following main problems when facing complex interferometric structures (such as OAM composite models) and the need for high-precision decoupling of multiple ocean parameters:

[0013] (1) Relying on one-dimensional angular dimensionality reduction results in the loss of spatial information:

[0014] Polar coordinate transformation and angular intensity extraction ignore the rich two-dimensional structural information in the interferogram. Especially when the image is noisy, eccentric, or locally blurred, the angular curve is difficult to accurately restore the overall rotation features.

[0015] (2) The main peak of the spectrum is severely leaked, and the interpolation estimation error is large:

[0016] Buneman et al.'s three-point interpolation method is only effective for the center of the ideal sinc model and cannot cope with problems such as leakage, asymmetry, and offset of the main lobe of the real spectrum, which leads to the estimation results deviating from the true value.

[0017] (3) It cannot adapt to complex composite patterns (such as concentric multi-ring structures):

[0018] When the interferogram contains multiple concentric OAM modes, composite interferometric rings, or asymmetric brightness distributions, traditional methods cannot distinguish the precise rotational shift of the dominant frequency peak, making it difficult to support simultaneous identification of multiple parameters.

[0019] (4) Cannot handle main lobe angle drift in two-dimensional spectra:

[0020] Current methods generally lack the ability to fit the main frequency peak at the subpixel level in the two-dimensional frequency domain, and cannot directly invert the perturbation angle from the image itself, which restricts the overall accuracy and robustness of the system.

[0021] Existing patents also include some measurement devices and / or methods related to optical fiber sensing demodulation based on OAM (orbital angular momentum), such as the Chinese invention patent application CN202310221113.3 entitled "Optical Orbital Angular Momentum Interference Demodulation Technology Based on Four-Point Angular Probe." This patent measures light intensity at any angular detection point in a petal-shaped interference pattern, and the interference phase change can be calculated after normalization. Because it uses four photodetectors instead of the traditional image acquisition and processing methods for orbital angular momentum interference signal demodulation, and only requires four detection points in any different angular directions on an interference pattern to correctly demodulate the phase, the demodulation process is greatly simplified. However, it is mainly used to solve the problems of insufficient detection resolution and low speed in measuring small optical path (phase) changes, and its measurement method requires the use of Fourier transform and shift correlation to determine the angular angle.

[0022] For example, the Chinese invention patent with application number CN200910086930.2, entitled "Method and Apparatus for Demodulating the Orbital Angular Momentum State of a Hybrid Helical Phase Beam," can detect the orbital angular momentum state and energy distribution of each helical component contained in a hybrid helical phase beam. However, it uses an energy meter to measure the reflected energy to determine the energy distribution of the helical components in the beam under test, and uses the transmitted light from two beam splitters at the two output ends of an interferometer to be incident on a diffraction grating, and determines the orbital angular momentum state of the output light at the output end by observing the distribution of each diffraction order. It does not involve improving demodulation accuracy or demodulation efficiency. Summary of the Invention

[0023] To overcome the shortcomings of existing technologies, the present invention aims to provide a high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm, which is suitable for large dynamic range interferometry. It has high demodulation accuracy, sub-angle level demodulation capability, and accuracy and linear stability under multi-angle conditions. Its process is simple and computationally complex.

[0024] This invention employs the following technical solution: a high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm, comprising the following steps:

[0025] Step 1: Obtain the interference image; use simulation or experimental methods to obtain the two-dimensional original image formed by the interference of the Gaussian beam and the OAM beam; the center of the two-dimensional original image has an obvious symmetrical structure, a chrysanthemum-shaped pattern, and periodically distributed bright spots; the image input is a grayscale image I(x,y) with a size of N×N, and the grayscale value represents the light intensity;

[0026] Step 2: Preprocess the image; preprocess the input two-dimensional original image I(x,y) by reducing noise through Gaussian filtering, normalizing pixel values ​​to [0,1], and removing background noise through intensity thresholding to enhance the recognizability of bright spot signals and improve the reliability of extreme point extraction. The preprocessed image is denoted as I3(x,y).

[0027] Step 3: Extract the extreme points of the bright spot; call the two-dimensional region extreme point detection algorithm to compare all pixels in the preprocessed image I3(x,y) with their neighborhood, extract the local maximum points, and denote the initial set of bright spot center coordinates as P;

[0028] Step 4: Perform sub-pixel-level center localization; for the extracted initial bright spot center coordinate set P={(x i ,y i Local interpolation is performed on each point in the matrix to obtain a more accurate sub-pixel center position, whose coordinates are denoted as (x...). i ′,y i ′);

[0029] Step 5: Perform polar coordinate transformation on the sub-pixel-level center coordinates; for each obtained bright spot, calculate the sub-pixel-level center coordinates (x... i ′,y i ′), with the image center (x c ,y c Using point (x, y) as a reference, convert to polar coordinates to obtain the polar radius r of each bright spot. i and polar angle θ i ,

[0030] The resulting set of bright spot polar angles is represented as follows:

[0031] ;

[0032] Step 6: Rotation angle fitting and estimation.

[0033] The theoretical bright spot angle sequence is constructed as follows:

[0034] θ ideal =(2π*k) / N, k=0,1,...,N-1

[0035] The actual detected bright spot polar angle sequence θ i Sort in ascending order, and compare with the above θ ideal By performing a one-to-one correspondence, a preliminary estimate of the average rotation angle shift Δθ between the actual interferogram and the theoretical pattern is obtained:

[0036] Δθ=mean(θ i -θ ideal );

[0037] Where mean() represents taking the arithmetic mean, which means taking the polar angle sequence θ of all actually detected bright spots. i With the corresponding theoretical bright spot angle sequence θ ideal The differences are calculated sequentially and their arithmetic mean is taken to estimate the average offset of the overall rotation angle;

[0038] Step 7: Output the average rotation angle offset Δθ as the interferogram rotation angle corresponding to the phase difference between the OAM beam and the Gaussian beam. Convert it to degrees as required, in degrees. At the same time, output intermediate variables such as bright spot polar coordinates and center position for debugging and verification.

[0039] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0040] 1. The present invention discloses a high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm. The theoretical bright spot angle sequence obtained by the method almost coincides with the actual detected bright spot polar angle sequence, resulting in high demodulation accuracy. By extracting the extreme points of the bright spot and achieving sub-pixel-level center positioning, the method enables high-precision angle demodulation even under conditions of limited interferometric image resolution. The estimated rotation angle is highly consistent with the theoretical rotation angle, with an average error of less than 1°. It has sub-angle-level demodulation capability, accuracy and linear stability under multi-angle conditions, and is suitable for large dynamic range interferometric measurements.

[0041] 2. The present invention provides a high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm. This method does not require frequency domain transformation or frequency domain analysis, and is entirely based on two-dimensional image space operations. It extracts multiple bright spot center points in the OAM interference pattern through image grayscale intensity analysis, avoiding reliance on complex tools such as Fourier transform and spectrum analysis. It has higher computational efficiency and lower computational complexity, making it suitable for integration into low-computing-power platforms. It does not require a stable laser interference frequency or high-speed sampling equipment, and can adapt to rapid interferogram processing under changing conditions. It is suitable for applications such as online demodulation and dynamic tracking. Attached Figure Description

[0042] Figure 1 This is a flowchart of the high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm of the present invention.

[0043] Figure 2a This is a two-dimensional spatial intensity distribution diagram of the Gaussian beam in Example 1; Figure 2b The intensity distribution diagram of the OAM beam; Figure 2c The original two-dimensional image formed by the interference of a Gaussian beam and an OAM beam; Figure 2d The phase distribution diagram of the Gaussian beam; Figure 2e The phase distribution diagram of the OAM beam; Figure 2f The two-dimensional fast Fourier transform of the interference pattern;

[0044] Figure 3 It is the initial bright spot image identified in Example 1;

[0045] Figure 4 It is the pixel grayscale value of the center point of an initial bright spot obtained by using the 3×3 neighborhood window comparison method in Example 1;

[0046] Figure 5 This is a distribution diagram comparing the theoretical bright spot angle sequence and the actual detected bright spot polar angle sequence in polar coordinates in Example 1.

[0047] Figure 6 This is a comparison diagram of the input phase difference and demodulation angle in Example 1. Detailed Implementation

[0048] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0049] The purpose of this invention is to address the shortcomings of existing technologies by providing augmented reality glasses and an image processing method thereof.

[0050] Example 1

[0051] A high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm, referring to Figure 1 As shown, it includes the following steps:

[0052] Step 1: Obtain the interference image. Use simulation or experimental methods to obtain the two-dimensional original image formed by the interference of the Gaussian beam and the OAM (orbital angular momentum) beam; the center of the two-dimensional original image has a clear symmetrical structure, with a chrysanthemum-like pattern and periodically distributed bright spots; the image input is a grayscale image I(x,y) with a size of N×N, and the grayscale value represents the light intensity.

[0053] Reference Figure 2a The image shows the two-dimensional spatial intensity distribution of the Gaussian beam used in this embodiment: the standard height is high, with the highest brightness at the center and decreasing radially outwards. (Refer to...) Figure 2d As shown, this is the phase distribution of the Gaussian beam. The phase is uniform and constant, which characterizes its plane wavefront properties.

[0054] Reference Figure 2b The image shows the intensity distribution of the OAM beam used in this embodiment. Its topological charge number is l=6, and the intensity exhibits a distinct ring-shaped distribution with a phase singularity at the center. (Refer to...) Figure 2e As shown, the phase distribution of the OAM beam is displayed, showing the OAM mode with a spiral phase structure. The color cycles 6 times from 0 to ±π, presenting a typical angular phase change.

[0055] Reference Figure 2c The image shown is a two-dimensional original image formed by the superposition of the interference of a Gaussian beam and an OAM beam. It exhibits six bright spots, representing the topological charge of the OAM beam modes. The image center shows obvious symmetry, and the six bright spots, distributed in a ring, form a "chrysanthemum-shaped" interference fringe. (Refer to...) Figure 2f As shown, the result of the two-dimensional fast Fourier transform of the interferogram highlights the OAM mode information in the interferogram due to its symmetry.

[0056] Step 2: Image preprocessing. The input two-dimensional original image I(x,y) is preprocessed by reducing noise through Gaussian filtering, normalizing pixel values ​​to [0,1], and removing background noise through intensity thresholding to enhance the recognizability of bright spot signals and improve the reliability of extreme point extraction. The preprocessed image is denoted as I3(x,y).

[0057] Specifically, the image preprocessing in step 2 includes the following steps:

[0058] Step 21: Use a two-dimensional Gaussian kernel to perform Gaussian filtering noise reduction on the input two-dimensional original image I(x,y). Set the filter standard deviation σ to 1~2 and the kernel size to 5×5 or 7×7. The filtered image is denoted as I1(x,y). This step is used to suppress background noise and isolated noise points and enhance the intensity continuity of local regions.

[0059] Step 22: Perform linear normalization on the filtered image, mapping all pixel values ​​to the [0,1] interval. The calculation method is as follows:

[0060]

[0061] Step 23: Set a gray intensity threshold for the normalized image I2(x,y). In this embodiment, the gray intensity threshold is 0.3. Pixels below this threshold are set to zero or ignored to remove non-interference bright spot areas and improve the reliability of extreme point extraction. The processed image is denoted as I3(x,y).

[0062] Step 24: Perform a masking operation on the N-pixel width of the edges around image I3(x,y) and force it to be 0 to ensure that features are extracted only within the effective area of ​​the image and avoid false detections of boundaries.

[0063] Step 3: Extract the extreme points of the bright spot; call the two-dimensional region extreme point detection algorithm to compare all pixels in the preprocessed image I3(x,y) with their neighborhood, extract the local maximum point, and denote the initial set of bright spot center coordinates as P.

[0064] Specifically, step 3, extracting the extreme points of the bright spot, includes the following steps:

[0065] Step 31: Employ a regional extremum search algorithm. Using the `imregionalmax` function in MATLAB and / or a 3×3 neighborhood window comparison method, traverse all pixels in image I3(x,y) and compare them with their neighboring pixel values. If the current pixel value is greater than all its neighboring pixels and greater than the set extremum intensity threshold, then the point is considered a valid bright spot center. In this embodiment, the extremum intensity threshold is 0.4. The `imregionalmax` function is used to detect the local maximum position in each connected region of a two-dimensional image or matrix. Its principle is to compare the gray values ​​of the target pixel with those of its neighboring pixels, retain only the points that are greater than or equal to all neighboring pixel values, and output the corresponding logical mask matrix.

[0066] Step 32: Calculate the image coordinates (x, y) of all points that meet the conditions. i ,y i Store the initial set of bright spot center coordinates:

[0067]

[0068] Step 33: When the brightness density is too high or the centers of adjacent brightness spots are too close, candidate points are screened by setting the minimum spacing or response sorting to avoid multiple detections of the same brightness spot area; the extracted initial brightness spot center coordinates are used for sub-pixel accuracy estimation.

[0069] Reference Figure 3 As shown, the pixel-level coordinates of the initial bright spot center are identified using the `imregionalmax` function in MATLAB through a regional extremum point search algorithm. The bright spot peaks are marked with white outlines, representing local bright spot intensity extrema in the image. By performing bright spot extremum point extraction on the interferometric image after Gaussian filtering and intensity thresholding preprocessing, bright spot peaks with centrosymmetric characteristics can be effectively extracted, and pixel-level initial coordinates of the bright spot center can be constructed. This extraction method has stronger noise resistance and higher local judgment accuracy compared to traditional maximum intensity methods or simple centroid methods.

[0070] Step 4: Perform sub-pixel-level center localization; for the extracted initial bright spot center coordinate set P={(x i ,y iLocal interpolation is performed on each point in the matrix to obtain a more accurate sub-pixel center position, whose coordinates are denoted as (x...). i ′,y i ′);

[0071] Specifically, step 4, subpixel-level center localization, includes the following steps:

[0072] Step 41: Perform local interpolation calculations for each point in the extracted initial bright spot center coordinate set P, using the current center pixel (x... i ,y i Based on this, extract the grayscale values ​​of its left, center, and right neighboring pixels:

[0073]

[0074]

[0075]

[0076] Calculate the sub-pixel offset in the horizontal direction:

[0077]

[0078] Step 42: Extract the values ​​of its upper and lower neighboring pixels:

[0079]

[0080]

[0081] Calculate the vertical sub-pixel offset:

[0082]

[0083] Step 43: If the denominator in the above calculation is close to 0, such as less than the set threshold ε, it indicates that the local gray level is symmetrical or the noise influence is too large. In this case, the point can be regarded as an invalid interpolation point, and you can choose to skip it or use other interpolation methods such as quadratic fitting to replace it.

[0084] Step 44: Correct the initial bright spot center coordinates based on the calculated horizontal and vertical sub-pixel offsets:

[0085]

[0086] Where i = 1, 2, 3, ..., M, the center positions of all sub-pixels are obtained. This will be used for subsequent polar coordinate transformation and angle demodulation operations.

[0087] Reference Figure 4As shown, the intensity values ​​of a certain initial bright spot center point obtained by the regional extreme point search algorithm and the comparison method of a 3×3 neighborhood window are extracted from its top, bottom, left, right and center pixels, totaling 3×3 pixels. The values ​​marked in the figure are the normalized pixel gray values. The nine pixel gray values ​​from left to right and from top to bottom are 0.17, 0.11, 0.07, 0.16, 0.11, 0.06, 0.15, 0.10 and 0.06, which are used to calculate the sub-pixel offsets in the horizontal and vertical directions.

[0088] In step 4, the intensity values ​​of three adjacent pixels are extracted along the horizontal and vertical directions for each initial extreme point, and the sub-pixel offset is calculated by applying the interpolation formula to obtain a more accurate bright spot center position. Compared with the three-point interpolation or Gaussian fitting methods commonly used in the prior art, the interpolation method of the present invention has the characteristics of structural symmetry, no iteration required, and small computational load. It is suitable for processing interferogram data with small bright spot size, limited image resolution, or strong noise background. For example, with a 3×3 pixel window, even if the bright spot coverage area is extremely small, its center position can still be accurately recovered by interpolation.

[0089] Step 5: Perform polar coordinate transformation on the sub-pixel-level center coordinates; for each obtained bright spot, calculate the sub-pixel-level center coordinates (x... i ′,y i ′), with the image center (x c ,y c Using point (x, y) as a reference, convert to polar coordinates to obtain the polar radius r of each bright spot. i and polar angle θ i ,

[0090] Specifically, the polar coordinate transformation relationship for sub-pixel level center coordinates is as follows:

[0091]

[0092] ,

[0093] The atan2 function is used to calculate the polar angle of the bright spot relative to the center, which can accurately distinguish the four quadrants; all angle values ​​are uniformly converted to the interval [0, 2π) to ensure the continuity and consistency of the angle sequence; , .

[0094] The resulting set of bright spot polar angles is represented as follows:

[0095]

[0096] Step 6: Fitting and estimating the rotation angle;

[0097] When a Gaussian beam interferes with an OAM beam, multiple periodic bright spots appear in the resulting interference pattern, and their number is positively correlated with the modulus of the OAM. Based on the spiral interference mechanism of the interference fringes, if the topological charge of the OAM beam is L, then the theoretical number of bright spots formed by its interference with the Gaussian mode is:

[0098] N=L

[0099] This pattern originates from the complex amplitude term e. iLθ When superimposed with the angular phase during fundamental mode interference, it introduces L complete spiral cycles within one 2π revolution, thus forming L bright spots.

[0100] Based on this, the theoretical bright spot angle sequence is constructed as follows:

[0101] θ ideal =(2π*k) / N, k=0,1,...,N-1

[0102] The actual detected bright spot polar angle sequence θ i Sort in ascending order, and compare with the theoretical bright spot angle sequence θ mentioned above. ideal By performing a one-to-one correspondence, a preliminary estimate of the average rotation angle shift Δθ between the actual interferogram and the theoretical pattern is obtained:

[0103] Δθ=mean(θ i -θ ideal )

[0104] Where mean() represents taking the arithmetic mean, which means taking the polar angle sequence θ of all actually detected bright spots. i With the corresponding theoretical bright spot angle sequence θ ideal The differences are calculated sequentially, and their arithmetic mean is taken to estimate the average offset of the overall rotation angle.

[0105] Reference Figure 5 As shown, the theoretical bright spot angle sequence θ is compared in polar coordinates. ideal Compared with the actual detected bright spot polar angle sequence θ i The blue dashed lines represent the theoretical bright spot angle sequence θ. ideal The red dots and broken lines represent the actual detected bright spot polar angle sequence θ. i The radial scale (0, 0.5, 1, 1.5, 2) in the figure represents a unit radius, used for visualization, and has no physical dimensions. The theoretical bright spot angle sequence θ obtained in this embodiment... ideal Compared with the actual detected bright spot polar angle sequence θ i The near-overlapping values ​​indicate that the high-dimensional OAM mode sensing demodulation method of this invention has high demodulation accuracy.

[0106] Step 7: The output average rotation angle offset Δθ is used as the interferogram rotation angle corresponding to the phase difference between the OAM beam and the Gaussian beam. It is converted to the angle system as required, in degrees. At the same time, intermediate variables such as bright spot polar coordinates and center position are output for debugging and verification.

[0107] Reference Figure 6 As shown, this is a comparison chart of input phase difference and demodulation angle: This chart displays the rotation angle results output by the demodulation method under different input phase differences (rotation angles). The horizontal axis represents the set theoretical rotation angle, i.e., the relative phase difference introduced when the OAM beam interferes with the Gaussian beam, in degrees. The vertical axis represents the image rotation angle estimated by the demodulation method of this invention, in degrees. The red dots represent the demodulation output values, and the black dashed lines represent the theoretical one-to-one correspondence reference lines. The results show that the estimated rotation angle and the theoretical rotation angle are in high agreement, with an average error of less than 1°. This indicates that the high-dimensional OAM mode sensing demodulation method of this invention has sub-angle-level demodulation capability, verifying the accuracy and linear stability of the method under multi-angle conditions, and demonstrating good repeatability and robustness. Within the entire ±60° input range, the demodulation results are almost strictly linear, verifying its applicability to large dynamic range interferometry. This invention does not use frequency domain analysis, yet still exhibits high stability even with low image resolution. Based solely on image spatial processing, the method is simple and computationally inexpensive. This angle demodulation method, which does not rely on frequency domain transformation and is entirely based on image domain operations, greatly simplifies system implementation while maintaining high accuracy. It provides reliable support for the promotion of OAM angle measurement in resource-constrained, embedded, or dynamic application scenarios, and is suitable for integration into low-computing-power platforms. It does not require a stable laser interference frequency or high-speed sampling equipment, can adapt to rapid interferogram processing under changing conditions, and is suitable for online demodulation, dynamic tracking, and other application scenarios.

[0108] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm, characterized in that: Includes the following steps, Step 1: Obtain the original two-dimensional image I(x,y) formed by the interference of the Gaussian beam and the OAM beam; Step 2: Apply Gaussian filtering to I(x,y), normalize the pixel values ​​to [0,1], and remove background noise by intensity thresholding. The preprocessed image is denoted as I3(x,y). Step 3: Use the two-dimensional region extreme point detection algorithm to compare all pixels in image I3(x,y) with their neighborhood, extract the local maximum points, and obtain the initial set of bright spot center coordinates; Step 4: Perform sub-pixel-level center localization; for each point (x, y) in the initial set of bright spot center coordinates... i ,y i Local interpolation calculations are performed to obtain a more accurate sub-pixel center position (x). i ′,y i ′); Step 5: Perform polar coordinate transformation; for each bright spot, the sub-pixel center (x i ′,y i (x') with the image center (x) c ,y c Using point (x, y) as a reference, convert to polar coordinates to obtain the polar radius r of each bright spot. i and polar angle θ i The set of bright spot polar angles is represented as: ； Step 6: Rotation angle fitting and estimation, constructing the theoretical bright spot angle sequence as follows: θ ideal =(2π*k) / N,k=0,1,...,N-1 The actual detected bright spot polar angle sequence θ i Sort in ascending order, and compare with the above θ ideal By performing a one-to-one correspondence, a preliminary estimate of the average rotation angle shift Δθ between the actual interferogram and the theoretical pattern is obtained: Δθ=mean(θ i -θ ideal ); The mean() function represents taking the arithmetic mean, which is calculated sequentially and then taken as the arithmetic mean to estimate the average offset of the overall rotation angle.

2. The high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm according to claim 1, characterized in that: It also includes Step 7: Output the average rotation angle offset Δθ as the interferogram rotation angle corresponding to the phase difference between the OAM beam and the Gaussian beam, convert it to degrees in °, and output the polar coordinates and center position of the bright spot for debugging and verification.

3. The high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm according to claim 1 or 2, characterized in that: Step 2, image preprocessing, includes the following steps: Step 21: Use a two-dimensional Gaussian kernel to perform Gaussian filtering on the input two-dimensional original image I(x,y) for noise reduction. The filtered image is denoised as I1(x,y). Step 22: Perform linear normalization on the filtered image, mapping all pixel values ​​to the [0,1] interval. The calculation method is as follows: Step 23: Set a gray intensity threshold for image I2(x,y) to remove non-interference bright spot areas. The processed image is denoted as I3(x,y). Step 24: Perform a masking operation on the edges of image I3(x,y) with a certain pixel width to ensure that features are extracted only within the effective area of ​​the image.

4. The high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm according to claim 3, characterized in that: Step 3, extracting the extreme points of the bright spot, includes the following steps: Step 31: Use the regional extreme point search algorithm to traverse all pixels in image I3(x,y) and compare them with the values ​​of its neighboring pixels. If the current pixel value is greater than all its neighboring pixels and greater than the set extreme intensity threshold, then the point is regarded as the center of the effective bright spot. Step 32: Calculate the image coordinates (x, y) of all points that meet the conditions. i ,y i Store the initial set of bright spot center coordinates; Step 33: Set minimum spacing or response sorting to filter candidate points, and use the extracted initial bright spot center coordinates for sub-pixel accuracy estimation.

5. The high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm according to claim 4, characterized in that: Step 4, subpixel-level center localization, includes the following steps: Step 41: Perform local interpolation calculations for each point in the extracted initial set of bright spot center coordinates, using the current center pixel (x... i ,y i Based on this, extract the grayscale values ​​of its left, center, and right neighboring pixels: Calculate the sub-pixel offset in the horizontal direction: Step 42: Extract the values ​​of its upper and lower neighboring pixels: Calculate the vertical sub-pixel offset: Step 43: If the denominator in the above calculation is close to 0, then the point is considered an invalid interpolation point; Step 44: Correct the initial bright spot center coordinates based on the calculated horizontal and vertical sub-pixel offsets: Where i = 1, 2, 3, ..., M; the center positions of all sub-pixels obtained. Used for subsequent polar coordinate transformation and angle demodulation operations.

6. The high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm according to claim 5, characterized in that: The polar coordinate transformation relationship for the sub-pixel level center coordinates in step 5 is as follows: ， ， The atan2 function is used to calculate the polar angle of the bright spot relative to the center, and all angle values ​​are uniformly converted to the interval [0, 2π).

7. The high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm according to claim 3, characterized in that: In step 23, the gray intensity threshold set for the normalized image I2(x,y) is 0.2~0.

4.

8. The high-dimensional OAM mode fiber optic sensing demodulation method based on the Buneman algorithm according to claim 4, characterized in that: In step 31, the imregionalmax function in MATLAB or the method of comparing all pixels of image I3(x,y) using a 3×3 neighborhood window is used.

Images