Self-adaptive free space light tracking and pointing system and method

Abstract

The invention discloses a self-adaptive free space light tracking and pointing system and method. The system comprises but is not limited to a CMOS camera, a grayscale image module, a signal light mode field, an optical fiber mode field and a controller. The method comprises the following steps: calibrating an optical fiber mode field, generating a fixed-size convolution kernel and storing calibration data; capturing a light spot image and preprocessing to output a stable light spot image; dynamically searching a light spot region to output an effective region coordinate; calculating a convolution matching degree to generate a power matrix; and feedback control, adjustment and driving of the optical execution mechanism. According to the method, the optimal power point is accurately calculated according to tracking and pointing of the turbulence self-adaptive spatial light, a power point calculation mode of a traditional centroid algorithm is replaced, the influence of complex environmental factors such as atmospheric turbulence and system distortion on the transmission efficiency is effectively reduced, and therefore efficient coupling transmission of the signal light is achieved.

Description

Technical Field

[0001] The present invention relates to the field of free-space optical communication, and particularly to an adaptive free-space optical acquisition, tracking and pointing system and method. Background Art

[0002] Free-Space Optical Communication (FSO) is a communication technology based on the propagation of optical signals in free space, which has advantages such as high bandwidth, high speed, and anti-electromagnetic interference, and is widely used in scenarios such as satellite communication, unmanned aerial vehicle communication, and ground high-speed communication.

[0003] As the core control unit of free-space optical communication, the Acquisition, Tracking, and Pointing System (ATP system), its dynamic accuracy directly determines the optical signal coupling efficiency. Traditional ATP systems usually rely on the intensity distribution characteristics of the beacon light, obtain the spot position information in real time through a high-speed photodetector, and drive the Fast Steering Mirror (FSM) to complete the closed-loop adjustment of the light beam.

[0004] In the prior art, to achieve efficient coupling of the signal light and the optical fiber, the centroid method of the beacon light is usually used, and the centroid of the beacon light spot is used as the incident point of the signal light, and the best match with the optical fiber mode field is achieved by adjusting the direction of the signal light. However, in a dynamic turbulence scenario, the ATP system faces two key bottlenecks: First, the non-uniformity of the transmission paths of the beacon light and the signal light due to the wavelength difference makes it difficult for centroid tracking to reflect the true offset of the signal light mode field; Second, the existing closed-loop control relies on the positioning of the geometric center of the light spot and does not consider the spatial overlap characteristics of the optical fiber mode field and the incident light field, resulting in a lag in dynamic error compensation. Summary of the Invention

[0005] Object of the Invention: The object of the present invention is to provide an adaptive free-space optical acquisition, tracking and pointing system and method.

[0006] Technical Solution: The adaptive free-space optical acquisition, tracking and pointing system described in the present invention includes a CMOS camera, a grayscale image module, a signal light mode field, an optical fiber mode field, and a controller. The CMOS camera acquires an image, extracts the spot part according to the acquired image and calibrates the signal light mode field through intensity information. The mode field information of the optical fiber is calibrated in advance in the laboratory. The efficiency distribution is obtained after convolution of the image and the optical fiber mode field. The power optimal point is obtained according to the position of the maximum value in the efficiency distribution. Finally, the controller controls the movement of the light spot.

[0007] Further, the controller includes a voice coil motor and a piezoelectric ceramic.

[0008] Further, the grayscale image module is used to extract grayscale values from the acquired image and represent the light intensity through the grayscale values.

[0009] The adaptive free-space optical tracking method of the present invention includes the following steps:

[0010] (1) Fiber optic mode field calibration and generation and storage of calibration data with a fixed-size convolution kernel. Obtain fiber optic mode field data through laboratory calibration and preprocess it into a fixed-size convolution kernel;

[0011] (2) Spot image capture and preprocessing to output a stable spot image. Capture the data in the CMOS camera and perform two-level processing of dynamic weighted fusion - adaptive filtering to significantly improve the quality of the spot image in a turbulent environment;

[0012] (3) Dynamic search for the spot area to output the coordinates of the effective area. Rapidly locate the effective area of the spot by iteratively adjusting the position and size of the search box;

[0013] (4) Calculate the convolution matching degree to generate a power matrix. Calculate the matching degree distribution of the spot and the mode field through direct spatial domain convolution;

[0014] (5) Locate the global optimal point to output the image coordinate deviation. Determine the best matching position by analyzing the power matrix;

[0015] (6) Feedback control adjustment to drive the optical actuator. Generate a control signal through the image coordinate deviation and dynamically adjust the optical path alignment.

[0016] Further, the step (1) includes:

[0017] (1.1) Calibration of the mode field intensity distribution

[0018] Under the illumination of a standard light source, use a microscopic imaging system to measure the light intensity distribution I fiber (x, y) of the output end face of the optical fiber;

[0019] (1.2) Gaussian fitting simplification

[0020] For a single-mode optical fiber, fit the measurement data to a two-dimensional Gaussian function:

[0021]

[0022] where w0 is the mode field diameter, and retain I0 and w0 as key parameters, x represents, and y represents the horizontal / vertical coordinates in the image;

[0023] (1.3) Generation of a fixed-size convolution kernel

[0024] Generate a normalized convolution kernel of N×N pixels

[0025]

[0026] Among them Ensure that ∑K 2 (x, y) = 1, where K represents the normalized mode field convolution kernel.

[0027] The formula for calculating the mode field overlap degree is as follows:

[0028]

[0029] Among them, E1 represents the mode field distribution at the output end face of the optical fiber, and E2 represents the mode field distribution captured by the CMOS.

[0030] Furthermore, step (2) includes:

[0031] (2.1) Signal-to-noise ratio definition and weight assignment

[0032] The single-frame signal-to-noise ratio is defined as:

[0033]

[0034] where μ ROI，k represents the pixel mean of the estimated spot area (based on historical data), and σ bg,k represents the standard deviation of the image edge background area.

[0035] The exponential weight assignment is:

[0036]

[0037] This processing weight distribution satisfies

[0038] (2.2) Multi-frame weighted fusion

[0039] Perform a weighting operation on the image data, and the weighting formula is as follows:

[0040]

[0041] where is the original image data of the k-th frame, and w k represents the weighting value at time k. represents the spot value at (x, y) in the original image at time k, and I fused (x, y) represents the finally used image data.

[0042] Furthermore, step (3) includes:

[0043] (3.1) Initial search box setting

[0044] The default parameter configuration is the initial search box size: L0·L0, and the maximum expansion level: Nmax =3, initial position of center point:

[0045] (3.2) Determination of light spot existence

[0046] In the current search n Internal calculation of spot energy ratio:

[0047]

[0048] If the conditions are met, it is determined that the light spot exists and step 3.3 is executed; if the conditions are not met, it is determined that the light spot is lost and step (3.4) is executed;

[0049] (3.3) Fixed frame area capture

[0050] The most recent valid center Cut L0·L0 area

[0051]

[0052] (3.4) Expand the search box to scan

[0053] By level k=1, 2, 3...N max Gradually expand the search box:

[0054] L k =2 k 10

[0055]

[0056] The spot recapture condition is in the extended frame R k Internal detection energy density ρ k , if there exists ρ k ≥ρ th , then update the center point For R k The internal energy center of gravity:

[0057]

[0058] (3.5) Failure protection mechanism

[0059] If all expansion levels k≤N max If no light spot is detected, the image center is used as the starting point to perform a full image scan with a step length of L0 until ρ is found. k ≥ρ th If the global search still fails, the historical center point will be cleared and reset to the initial state.

[0060] Furthermore, the sliding convolution operation in step (4) includes:

[0061] Define the size of the convolution output matrix as (M - N + 1) × (M - N + 1).

[0062] Calculate the per - position matching degree. For each possible position (i, j) (0 ≤ i, j ≤ M - N) in the spot area, calculate:

[0063]

[0064] That is, the convolution kernel K slides on I ROI to calculate the local matching degree, where K(x, y) is the convolution kernel generated in step one.

[0065] Further, step (5) includes:

[0066] (5.1) Global maximum search

[0067] Matrix traversal:

[0068] (i max , j max ) = argmax P(i, j), 0 ≤ i ≤ W, 0 ≤ j ≤ H

[0069] Handle multi - peak conflicts. If there are multiple same maximum points, select the point whose geometric center is closest to the center of the matrix

[0070]

[0071] where i represents the row index of the power matrix, j represents the column index of the power matrix, w represents the width of the power matrix, and H represents the height of the power matrix;

[0072] (5.2) Global coordinate transformation

[0073] Combine the original image position of the windowing area R RoI to calculate the global coordinates

[0074] x global = x start + i max

[0075] y global = y start + j max

[0076] where x start represents the abscissa of the upper - left corner, and y start represents the ordinate of the upper - left corner.

[0077] Further, step (6) includes:

[0078] (6.1) Target position setting and definition of the deviation calculation reference coordinates

[0079] Define the center of the image as the theoretically optimal position

[0080] Calculate the pixel-level deviation between the current optimal position and the target:

[0081] Δx = x global - x ref and Δy = y global - y ref

[0082] where W image represents the total width of the image sensor, and H image represents the total height of the image sensor;

[0083] (6.2) PID control signal generation

[0084] Generate a piezoelectric ceramic drive voltage signal using proportional-integral-derivative control:

[0085]

[0086] where k p represents the proportional control coefficient, k i represents the integral control coefficient, and k d represents the derivative control coefficient;

[0087] (6.3) Optical actuator drive

[0088] Input the drive signals V x , V y into the two-dimensional piezoelectric ceramic translation stage to adjust the mirror angle and move the light spot towards the center of the image.

[0089] Beneficial effects: Compared with the prior art, the present invention has the following remarkable advantages: The present invention accurately calculates the power optimal point according to the turbulence adaptive spatial light tracking and aiming to replace the power point calculation method of the traditional centroid algorithm, effectively reducing the influence of complex environmental factors such as atmospheric turbulence and system distortion on the transmission efficiency, thereby realizing the efficient coupled transmission of the signal light. Brief description of the drawings

[0090] Figure 1 is a flowchart of the method of the present invention;

[0091] Figure 2 is a flowchart of the effective area coordinates of the dynamic search of the light spot area

[0092] Figure 3 is a structural diagram of the system of the present invention. Detailed implementation manners

[0093] The technical solution of the present invention will be further described below in conjunction with the drawings.

[0094] As Figure 1 shown, the adaptive free-space optical tracking method of the present invention includes the following steps:

[0095] (1) Fiber optic mode field calibration and generation and storage of calibration data for a fixed-size convolution kernel. Obtain fiber optic mode field data through laboratory calibration and preprocess it into a fixed-size convolution kernel;

[0096] (1.1) Mode field intensity distribution calibration

[0097] Under the illumination of a standard light source, use a microscopic imaging system to measure the light intensity distribution I fiber (x, y) of the output end face of the optical fiber;

[0098] (1.2) Gaussian fitting simplification

[0099] For a single-mode optical fiber, fit the measured data to a two-dimensional Gaussian function:

[0100]

[0101] where w0 is the mode field diameter, and I0 and w0 are retained as key parameters, x represents, and y represents the horizontal / vertical coordinates in the image;

[0102] (1.3) Generation of a fixed-size convolution kernel

[0103] Generate an N×N pixel normalized convolution kernel

[0104]

[0105] where Ensure that ∑K 2 (x, y) = 1, and K represents the normalized mode field convolution kernel.

[0106] The formula for the mode field overlap degree is as follows:

[0107]

[0108] where E1 represents the mode field distribution at the output end face of the optical fiber and E2 represents the mode field distribution captured by the CMOS.

[0109] (2) Spot image capture and preprocessing to output a stable spot image. Capture the data in the CMOS camera and perform secondary processing of dynamic weighted fusion - adaptive filtering to significantly improve the quality of the spot image in a turbulent environment;

[0110] (2.1) Definition of signal-to-noise ratio and weight assignment

[0111] The signal-to-noise ratio of a single frame is defined as:

[0112]

[0113] where μ ROI，k represents the pixel mean of the predicted spot area (based on historical data), and σ bg，k represents the standard deviation of the background area at the image edge.

[0114] The exponential weight distribution is as follows:

[0115]

[0116] This processing weight distribution satisfies

[0117] (2.2) Multi-frame weighted fusion

[0118] Perform a weighting operation on the image data, where the weighting formula is as follows:

[0119]

[0120] where is the original image data of the k-th frame, and w k represents the weighting value at time k represents the spot value at (x, y) in the original image at time k, and I fused (x, y) represents the final image data used.

[0121] (3) Dynamically search for the spot area and output the coordinates of the effective area, as Figure 2 shown, quickly locate the effective spot area by iteratively adjusting the position and size of the search box;

[0122] (3.1) Set the initial search box

[0123] The default parameter configuration is the initial search box size: L0·L0, the maximum expansion level: N max = 3, and the initial position of the center point:

[0124] (3.2) Determine the existence of the spot

[0125] Calculate the proportion of spot energy within the current search ρ n as follows:

[0126]

[0127] If satisfied, it is determined that the spot exists, and execute 3.3; if not satisfied, it is determined that the spot is lost, and execute step (3.4);

[0128] (3.3) Intercept the fixed frame area

[0129] Intercept the L0·L0 area with the most recent effective center ​

[0130]

[0131] (3.4) Extended search box scanning

[0132] By level k = 1, 2, 3... N max Gradually expand the search box:

[0133] L k = 2 k l0

[0134]

[0135] The spot re - capture condition is to detect the energy density ρ within the extended box R k inside, and if there exists ρ k such that k ≥ ρ th then update the center point to the inner energy center of gravity of R k :

[0136]

[0137] (3.5) Fail - safe mechanism

[0138] If no spot is detected for all expansion levels k ≤ N max then starting from the image center, perform a full - image scan with a step size of L0 until a region where ρ k ≥ ρ th is found. If it still fails after global search, clear the historical center point and reset to the initial state.

[0139] (4) Calculate the power matrix through convolution matching degree, and calculate the matching degree distribution of the spot and the mode field through direct - space - domain convolution;

[0140] Define the size of the convolution output matrix as (M - N + 1) × (M - N + 1),

[0141] Calculate the matching degree for each possible position (i, j) (0 ≤ i, j ≤ M - N) in the spot region:

[0142]

[0143] That is, the convolution kernel K slides on I ROI to calculate the local matching degree, where K(x, y) is the convolution kernel generated in step one.

[0144] (5) Locate the global optimal point and output the image coordinate deviation, and determine the best - matching position by analyzing the power matrix;

[0145] (5.1) Global maximum search

[0146] Matrix traversal:

[0147] (i max , j max ) = argmax P(i, j), 0 ≤ i ≤ W, 0 ≤ j ≤ H

[0148] Handling multi-peak conflicts. If there are multiple points with the same maximum value, select the point whose geometric center is closest to the center of the matrix

[0149]

[0150] Among them, i represents the row index of the power matrix, j represents the column index of the power matrix, w represents the width of the power matrix, and H represents the height of the power matrix;

[0151] (5.2) Global coordinate transformation

[0152] Combined with the original image position of the windowing area R ROI to calculate the global coordinates

[0153] x global = x start + i max

[0154] y global = y start + j max

[0155] Among them, x start represents the abscissa of the upper left corner, and y start represents the ordinate of the upper left corner.

[0156] (6) Feedback control adjusts the driving optical actuator to generate a control signal through the image coordinate deviation and dynamically adjusts the optical path alignment.

[0157] (6.1) Target position setting and deviation calculation reference coordinate definition

[0158] Define the image center as the theoretical optimal position

[0159] Calculate the pixel-level deviation between the current optimal position and the target:

[0160] Δx = x global - x ref , Δy = y global - y ref

[0161] Among them, W image represents the total width of the image sensor, and H image represents the total height of the image sensor;

[0162] (6.2) PID Control Signal Generation

[0163] Generate the piezoelectric ceramic drive voltage signal using proportional-integral-derivative control:

[0164]

[0165] Among them, k p represents the proportional control coefficient, k i represents the integral control coefficient, k d represents the derivative control coefficient;

[0166] (6.3) Optical Actuator Drive

[0167] Input the drive signals V x , V y into the two-dimensional piezoelectric ceramic translation stage to adjust the mirror angle and move the light spot towards the center of the image.

[0168] As Figure 3 shown, the adaptive free-space optical tracking system of the present invention includes a CMOS camera, a grayscale image module, a signal light mode field, a fiber mode field, and a controller. The CMOS camera acquires an image, extracts the light spot part based on the acquired image and through intensity information, calibrates the signal light mode field, calibrates the mode field information of the fiber in advance in the laboratory, obtains the efficiency distribution after convolution of the image and the fiber mode field, obtains the power optimal point based on the position of the maximum value in the efficiency distribution, and finally controls the movement of the light spot through the controller.

Claims

1. An adaptive free-space optical tracking and aiming system, characterized in that, It includes a CMOS camera, a grayscale image module, a signal light mode field, an optical fiber mode field, and a controller. The CMOS camera acquires an image, extracts the spot part according to the acquired image and through the intensity information, calibrates the signal light mode field, calibrates the mode field information of the optical fiber in advance in the laboratory, obtains the efficiency distribution after convolving the image and the optical fiber mode field, determines the power optimal point according to the position of the maximum value in the efficiency distribution, and finally controls the movement of the spot through the controller.

2. The adaptive free space optical tracking and aiming system according to claim 1, characterized in that The controller includes a voice coil motor and a piezoelectric ceramic.

3. The adaptive free space optical tracking and aiming system according to claim 1, characterized in that, The grayscale image module is used to extract the grayscale value from the acquired image and represent the light intensity through the grayscale value.

4. An adaptive free-space optical tracking and aiming method, characterized in that, Accurately calculate the spot mode field E by the mode field overlap principle A With the fiber mode field E B The matching degree of the optical path is dynamically adjusted, including the following steps: (1) Fiber mode field calibration and generation of a fixed-size convolution kernel, and storing the calibration data. The fiber mode field data E is obtained through laboratory calibration B , and it is preprocessed into a fixed-size convolution kernel; (2) Spot image capture and preprocessing output a stable spot image, capture the data in the CMOS camera, and significantly improve the quality of the spot image in a turbulent environment through two-level processing of dynamic weighted fusion - adaptive filtering; (3) Dynamic search for the spot area output the coordinates of the effective area, and quickly locate the effective area of the spot by iteratively adjusting the position and size of the search box; (4) Convolution matching degree calculation generates a power matrix, and calculates the matching degree distribution of the spot and the mode field through direct spatial domain convolution; (5) Global optimal point positioning output the image coordinate deviation, and determine the best matching position by analyzing the power matrix; (6) Feedback control adjusts the driving optical actuator, generates a control signal through the image coordinate deviation, and dynamically adjusts the optical path alignment.

5. The adaptive free-space optical tracking method according to claim 4, wherein The step (1) includes: (1.1) Calibration of the mode field intensity distribution Under the irradiation of a standard light source, use a microscopic imaging system to measure the light intensity distribution I of the output end face of the optical fiber fiber (x, y); (1.2) Gaussian fitting simplification For a single-mode optical fiber, fit the measurement data to a two-dimensional Gaussian function: where w0 is the mode field diameter, retain I0 and w0 as key parameters, x represents, and y represents the horizontal / vertical coordinate in the image; (1.3) Generation of a fixed-size convolution kernel Generate an N×N pixel normalized convolution kernel Among them Ensure that ∑K 2 (x, y) = 1, where K represents the normalized mode field convolution kernel. where the formula for calculating the mode field overlap degree is as follows: where, E1 represents the mode field distribution at the output end face of the optical fiber, and E2 represents the mode field distribution captured by the CMOS.

6. The adaptive free space optical tracking and aiming method according to claim 4, wherein The step (2) includes: (2.1) Definition of signal-to-noise ratio and weight assignment The signal-to-noise ratio of a single frame is defined as: where μ ROI，k represents the pixel mean of the predicted light spot area (based on historical data), and σ bg，k represents the standard deviation of the background area of the image edge. The exponential weight assignment is: The processing weight distribution satisfies (2.2) Multi-frame weighted fusion Perform a weighting operation on the image data, where the weighting formula is as follows: Among them is the original image data of the k-th frame, and w k represents the weighting value at time k represents the spot value of (x, y) in the original image at time k, and I fused (x, y) represents the finally used image data.

7. The adaptive free-space optical tracking and aiming method according to claim 4, wherein The step (3) includes: (3.1) Setting of the initial search box The default parameter configuration is the initial search box size: L0·L0, the maximum expansion level: N max = 3, the initial position of the center point: (3.2) Judgment of the existence of the spot In the current search ρ n calculate the proportion of the spot energy within: If satisfied, it is determined that the spot exists, and execute 3.3; if not satisfied, it is determined that the spot is lost, and execute step (3.4); (3.3) Interception of the fixed frame area With the most recent effective center Intercept the L0·L0 area (3.4) Scanning of the extended search box By level k = 1, 2, 3... N max Gradually expand the search box: L k = 2 k L0 The spot re-capture condition is to detect the energy density ρ within the expansion box R k If there exists ρ k ≥ρ k th then update the center point to the inner energy center of gravity of R k k : (3.5) Failure protection mechanism If all expansion series k ≤ N max If no light spot is detected, starting from the image center, perform a full-image scan with a step size of L0 until a region where ρ k ≥ ρ th is found. If the global search still fails, clear the historical center point and reset to the initial state.

8. The adaptive free-space optical tracking method according to claim 4, characterized in that The sliding convolution operation in the step (4) includes: Define the size of the convolution output matrix as (M-N+1)×(M-N+1), Calculate the matching degree for each possible position (i, j) (0≤i, j≤M-N) in the spot area one by one, and calculate: That is, the convolution kernel K slides on I ROI to calculate the local matching degree, where K(x, y) is the convolution kernel generated in Step 1.

9. The adaptive free space optical tracking method according to claim 4, wherein The step (5) includes: (5.1) Global maximum value search Matrix traversal: (i max , j max ) = argmax P(i, j), 0 ≤ i ≤ W, 0 ≤ j ≤ H Multi-peak conflict handling, if there are multiple identical maximum value points, select the point whose geometric center is closest to the center of the matrix where, i represents the row index of the power matrix, j represents the column index of the power matrix, w represents the width of the power matrix, and H represents the height of the power matrix; (5.2) Global coordinate transformation Combined with the original image position of the windowing area R RoI calculate the global coordinates x global = x start + i max y global = y start + j max Among them, x start represents the abscissa of the upper left corner, and y start represents the ordinate of the upper left corner.

10. The adaptive free-space optical tracking method according to claim 4, wherein The said step (6) includes: (6.1) Target position setting and deviation calculation reference coordinate definition Define the center of the image as the theoretically optimal position Calculate the pixel-level deviation between the current optimal position and the target: Δx = x global -x ref ,Δy = y global -y ref Among them, W image represents the total width of the image sensor, and H image represents the total height of the image sensor; (6.2) PID control signal generation Generate a piezoelectric ceramic drive voltage signal using proportional-integral-derivative control: Among them, k p represents the proportional control coefficient, k i represents the integral control coefficient, k d represents the derivative control coefficient; (6.3) Optical actuator drive Input the driving signals V x , V y into the two-dimensional piezoelectric ceramic translation stage, and adjust the mirror angle to move the light spot towards the center of the image.

Images