A method and system for tuning a fiber mode-locked laser based on image recognition algorithm

Abstract

This invention discloses a method and system for tuning fiber mode-locked lasers based on image recognition algorithms, relating to the field of fiber laser mode-locking technology. The method includes: preprocessing a spectrometer image; converting the spectral image from RGB color space to HSV color space, and separating the spectral curve from the image background through color filtering and selection; establishing a fully connected neural network (FNN) classification model based on linear image transformation to identify and classify the spectral curve into mode-locked and non-mode-locked states; determining the mode-locking intervals for different wavelengths by visualizing the PC angle intervals corresponding to the mode-locked states; traversing and searching within the specified wavelength mode-locking intervals, stopping PC rotation when a spectral image similar to the target image is detected. This intelligently tunes two different mode-locked states at different wavelengths.

Description

Technical Field

[0001] This invention relates to the field of fiber laser mode-locking technology, specifically a method and system for tuning fiber mode-locked lasers based on image recognition algorithms. Background Technology

[0002] The development and needs of fiber optic communication technology have spurred the development of multi-wavelength fiber lasers. This type of laser offers advantages such as a large number of output wavelengths, high beam quality, high conversion efficiency, and compact structure. Furthermore, in scientific and technological fields such as time-of-flight imaging and microscopy, Raman scattering spectroscopy, nonlinear frequency conversion, multicolor pump-probe spectroscopy, mid-infrared difference frequency generation, and differential radar, stable and close synchronization of optical pulses of different wavelengths is required. Therefore, dual-wavelength fiber lasers have attracted widespread attention and have been the subject of in-depth research and development.

[0003] Nonlinear polarization rotation (NPR) is a fiber laser structure used to generate and maintain soliton pulses, and it was the earliest and most widely used structure for soliton pulse fiber lasers. Due to the greater flexibility in wavelength and gain tuning range, NPR technology is also one of the important methods for realizing dual-wavelength fiber lasers. Compared to other mode-locking techniques, NPR fiber lasers have a simple structure, containing only two polarization controllers (PCs), and do not require complex components such as external modulators or gratings, making system design and manufacturing simpler. However, research on dual-wavelength NPR fiber lasers is still an emerging field. Under traditional conditions, mode-locking states at different wavelengths only occur under certain PC angle combinations. Manually adjusting the mode-locking of different wavelengths using NPR technology is highly random and can take several hours. Furthermore, after external environmental factors such as temperature and vibration cause the fiber laser to lose mode lock, a significant amount of time is required for readjustment.

[0004] Intelligent mode-locking methods based on machine learning or deep learning have become a research hotspot in the past two years. For example, K-means and LSTM have been successfully applied to automatic mode-locking of traditional soliton fiber lasers. However, current methods all use numerical inputs of spectral curves, which are time-consuming to train and are all aimed at single-wavelength traditional soliton mode-locking. There is still no intelligent mode-locking method that can be tuned to dual wavelengths. Summary of the Invention

[0005] To address the shortcomings mentioned in the background section, the present invention aims to provide a method and system for tuning fiber mode-locked laser based on an image recognition algorithm.

[0006] Firstly, the objective of this invention can be achieved through the following technical solution: a method for tuning fiber mode-locked laser based on an image recognition algorithm, the method comprising the following steps:

[0007] Receive spectrometer images, preprocess the spectrometer images to obtain preprocessed spectrometer images, generate image datasets from the preprocessed spectrometer images, and perform labeling operations on the image datasets to obtain labeled image datasets, wherein the spectrometer images are acquired based on the center wavelength set by the spectrometer.

[0008] The labeled image dataset is input into a pre-established FNN-based classification and recognition model for training, resulting in a trained FNN-based classification and recognition model. The spectrometer image to be identified is then obtained, input into the trained FNN-based classification and recognition model, and the image label is output.

[0009] Based on the image labels and the center wavelength set by the spectrometer, a scatter plot was used to draw the PC rotation angle combination of all images with the classification label as mode-locked. Data corresponding to different center wavelengths were marked with different colors, and the PC angle intervals corresponding to each center wavelength that showed the mode-locked state were obtained as the mode-locked intervals.

[0010] A center wavelength is specified, and the corresponding mode-locking interval is selected according to the specified center wavelength. The servo motor is controlled by the program to traverse the PC rotation angle combination in an orderly manner according to a fixed step size. When the target image appears, the PC rotation stops, thus realizing the tuning of fiber mode-locking.

[0011] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the acquisition process of the spectrometer image is as follows:

[0012] The center wavelength of the spectrometer is set to be near the center wavelength of the two types of solitons generated by the laser cavity. The servo motor is controlled by the program to rotate the PC in an orderly manner with a fixed step size, and a spectrometer image is acquired once for each rotation.

[0013] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the process of preprocessing the spectrometer image.

[0014] The input image is converted from its RGB color space spectrum to the HSV color space using the following formula:

[0015] V = mmax(R, G, B)

[0016]

[0017] In the formula, R, G, and B represent the values ​​of the red, green, and blue color components of an image, respectively, while H, S, and V represent the values ​​of the hue, saturation, and lightness components of an image, respectively. By setting upper and lower thresholds for the H, S, and V parameters, and using mask extraction, the image containing only the spectral curve is separated from the converted image.

[0018] By setting a rectangular convolution kernel and performing morphological operations such as opening, the separated image is filtered to remove noise.

[0019] Use Crop to crop the center of the image to obtain a square-sized image of the central portion;

[0020] The obtained image is converted into a grayscale image for output.

[0021] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the FNN based on the FNN classification and recognition model is a fully connected neural network based on image linear transformation, and the fully connected neural network based on image linear transformation includes: an input layer, a linear layer, four hidden layers, a softmax layer, and an output layer.

[0022] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the process of inputting the labeled image dataset into a pre-established FNN-based classification and recognition model for training.

[0023] The labeled image dataset is divided into a training set and a test set;

[0024] The fully connected neural network based on image linear transformation receives image data represented by a two-dimensional matrix in its input layer, and the linear layer transforms the two-dimensional matrix into a one-dimensional array which is then input into the hidden layer. Each neuron in each layer is fully connected, and the activation function is the ReLU function. The mapping relationship of the functions is as follows:

[0025]

[0026] Where x is the input of each neuron, and f(x) is the output after passing through the neuron;

[0027] The final activation function is the tanh function, and the mapping relationship of the functions is as follows:

[0028]

[0029] Where x is the input of each neuron, f(x) is the output after passing through the neuron; sinh(x) is the hyperbolic sine function, cosh(x) is the hyperbolic cosine function, and tanh(x) is the hyperbolic tangent function;

[0030] The output layer uses the softmax function to transform the output of the previous neural network into a probability value P between 0 and 1. i The formula is:

[0031]

[0032] Where z is a vector, representing the output of the previous layer of the neural network. i and z j It is one of the elements;

[0033] Set the loss function and optimizer. The loss function is the cross-entropy loss function, and the optimizer is the Adam algorithm.

[0034] The model is trained using the training set. After each training iteration, the training loss is calculated and tested on the test set. If the training loss converges and the average test accuracy exceeds a set threshold, the model training is stopped, and the trained FNN-based classification and recognition model is output.

[0035] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the process of controlling the servo motor to traverse the PC rotation angle combination in an orderly manner according to a fixed step size:

[0036] Initialize the servo angle and restore it to the vertical position;

[0037] Determine the required center wavelength and select the corresponding mode-locked interval as the input to the traversal control algorithm;

[0038] The servo motor is controlled by a traversal control algorithm to traverse the PC angle combination with a change step size of 1° to 1.5°. Each time the PC angle changes, the image output by the spectrometer is detected in real time.

[0039] If the output image is detected to be similar to the target image, the program terminates.

[0040] If the output image is found to be dissimilar to the target image, the detection continues until the program ends.

[0041] Secondly, in order to achieve the above objectives, this invention discloses a system for tuning fiber mode-locked lasers based on image recognition algorithms, comprising:

[0042] The image processing module is used to receive spectrometer images, preprocess the spectrometer images to obtain preprocessed spectrometer images, generate image datasets from the preprocessed spectrometer images, and perform labeling operations on the image datasets to obtain labeled image datasets. The spectrometer images are acquired based on the center wavelength set by the spectrometer.

[0043] The model training module is used to input the labeled image dataset into a pre-established FNN-based classification and recognition model for training, obtain the trained FNN-based classification and recognition model, acquire the spectrometer image to be identified, input the spectrometer image to be identified into the trained FNN-based classification and recognition model, and output the image label.

[0044] The interval determination module is used to draw a combination of PC rotation angles for all images with the classification label set by the spectrometer based on the image label and the center wavelength set by the spectrometer. The data corresponding to different center wavelengths are marked with different colors, and the PC angle intervals corresponding to each center wavelength that show the mode-locked state are obtained as the mode-locked intervals.

[0045] The mode-locked tuning module allows you to specify a center wavelength, select the corresponding mode-locked interval based on the specified center wavelength, and control the servo motor to traverse the PC rotation angle combination in an orderly manner with a fixed step size. When the target image appears, the PC rotation stops, thus achieving fiber mode-locked tuning.

[0046] In another aspect of the present invention, in order to achieve the above-mentioned objective, a terminal device is disclosed, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. The memory stores the computer program capable of running on the processor. When the processor loads and executes the computer program, it employs a method for tuning fiber mode-locked laser based on an image recognition algorithm as described above.

[0047] In another aspect of the present invention, in order to achieve the above-mentioned objective, a computer-readable storage medium is disclosed, wherein a computer program is stored in the computer program, and when the computer program is loaded and executed by a processor, a method for tuning fiber mode-locked laser based on an image recognition algorithm as described above is employed.

[0048] The beneficial effects of this invention are:

[0049] Based on the differences in spectral images under mode-locked and mode-free states, this invention constructs a recognition and classification model based on linear transformation. It innovatively finds all traditional soliton mode-locked intervals through image recognition. Based on this, soliton mode-locked images under two different bands can be automatically adjusted as needed.

[0050] When soliton mode-locking is achieved at different center bands, the present invention can determine the range of tunable center wavelength, that is, the range covering all center wavelengths.

[0051] This invention uses a common rotary servo motor, servo motor control board, Raspberry Pi, and thermoplastic to build a mechanical module for controlling a PC, achieving electric control of the PC at a very low cost and overcoming the problems of instability, randomness, and low precision of manual PC adjustment. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0054] Figure 2 This is a schematic diagram of a tunable dual-wavelength intelligent mode-locking system based on spectral image recognition provided in an embodiment of the present invention;

[0055] Figure 3 This is a schematic diagram of the image processing flow of the method of the present invention;

[0056] Figure 4 This is a neural network architecture diagram for image classification and recognition according to an embodiment of the present invention;

[0057] Figure 5 This is a schematic diagram of the wavelength tuning algorithm involved in the method of this invention;

[0058] Figure 6 This is a diagram showing the experimental results of dual-wavelength tuning in an embodiment of the present invention;

[0059] Figure 7 This is an example diagram of a conventional soliton boundary that can be formed near 1530nm within the laser cavity in an embodiment of the present invention;

[0060] Figure 8 This is an example diagram of a conventional soliton boundary that can be formed near 1560nm within the laser cavity in an embodiment of the present invention;

[0061] Figure 9 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Example 1:

[0064] The following is a description of the relevant terms used in the embodiments of this application:

[0065] Mode-locked fiber lasers based on the NPR (Neutral Resonance Phase Regulator) structure are a type of passively mode-locked fiber laser. Passive mode-locking refers to modulating the laser beam to obtain pulses using the saturable absorption characteristics of passive devices. Saturable absorbers are divided into naturally saturable absorbers and equivalent saturable absorbers. Commonly used naturally saturable absorbers include semiconductor saturable absorber mirrors or those based on materials such as carbon nanotubes and graphene. NPR technology utilizes a polarization controller (PC) and a polarization-dependent isolator to construct an equivalent saturable absorber, adjusting the self-phase modulation and cross-phase modulation of the pulse within the light beam to achieve mode-locking and obtain ultrashort pulses. Figure 1 As shown, a method for tuning fiber mode-locked laser based on image recognition algorithm includes the following steps:

[0066] S101: Receive spectrometer images, preprocess the spectrometer images to obtain preprocessed spectrometer images, generate image datasets from the preprocessed spectrometer images, and perform labeling operations on the image datasets to obtain labeled image datasets, wherein the spectrometer images are acquired based on the center wavelength set by the spectrometer.

[0067] Specifically, such as Figure 2 As shown, the hardware system of a tunable dual-wavelength intelligent mode-locking method based on spectral image recognition consists of three parts: a soliton fiber laser based on NPR structure and erbium-doped fiber, a servo control module, and a spectrometer. In the soliton fiber laser, the erbium-doped fiber (EDF) is 2.6m to 4m long, the single-mode fiber (SMF) is 15m to 20m long, and the pump (LD) power is 160mW to 200mW, which can achieve conventional soliton mode-locking near 1530nm and 1560nm. The servo control module includes a rotary servo, a mechanical structure made of thermoplastic plastic, a servo control board, a control board consisting of a Raspberry Pi and a computer, and the Raspberry Pi and the computer are connected via WIFI.

[0068] The process of acquiring spectrometer images is as follows:

[0069] By setting the center wavelength of the spectrometer to be near the center wavelength of the two types of solitons generated by the laser cavity, the PC is rotated in an orderly manner by the servo motor controlled by the program according to a fixed step size, and a spectrometer image is acquired once for each rotation.

[0070] The process of preprocessing spectrometer images:

[0071] like Figure 3 As shown, the input image is converted from the RGB color space spectrum to the HSV color space using the following formula:

[0072] V = mmax(R, G, B)

[0073]

[0074] In the formula, R, G, and B represent the values ​​of the red, green, and blue color components of an image, respectively, while H, S, and V represent the values ​​of the hue, saturation, and lightness components of an image, respectively. By setting upper and lower thresholds for the H, S, and V parameters, and using mask extraction, the image containing only the spectral curve is separated from the converted image.

[0075] By setting a rectangular convolution kernel and performing morphological operations such as opening, the separated image is filtered to remove noise.

[0076] Use Crop to crop the center of the image to obtain a square-sized image of the central portion;

[0077] The obtained image is converted into a grayscale image for output.

[0078] S102: Input the labeled image dataset into the pre-established FNN-based classification and recognition model for training to obtain the trained FNN-based classification and recognition model. Obtain the spectrometer image to be identified, input the spectrometer image to be identified into the trained FNN-based classification and recognition model, and output the image label.

[0079] The FNN-based classification and recognition model uses a fully connected neural network based on linear image transformations, such as... Figure 4 As shown, the fully connected neural network based on image linear transformation includes: an input layer, a linear layer, four hidden layers, a softmax layer, and an output layer.

[0080] The process of inputting the labeled image dataset into a pre-built FNN-based classification and recognition model for training:

[0081] The labeled image dataset is divided into a training set and a test set;

[0082] The fully connected neural network based on image linear transformation receives image data represented by a two-dimensional matrix in its input layer, and the linear layer transforms the two-dimensional matrix into a one-dimensional array which is then input into the hidden layer. Each neuron in each layer is fully connected, and the activation function is the ReLU function. The mapping relationship of the functions is as follows:

[0083]

[0084] Where x is the input of each neuron, and f(x) is the output after passing through the neuron;

[0085] The final activation function is the tanh function, and the mapping relationship of the functions is as follows:

[0086]

[0087] Where x is the input of each neuron, f(x) is the output after passing through the neuron; sinh(x) is the hyperbolic sine function, cosh(x) is the hyperbolic cosine function, and tanh(x) is the hyperbolic tangent function;

[0088] The output layer uses the softmax function to transform the output of the previous neural network into a probability value P between 0 and 1. i The formula is:

[0089]

[0090] Where z is a vector, representing the output of the previous layer of the neural network. i and z j It is one of the elements; in this embodiment, the classification and recognition model outputs only two results, so the z vector contains 2 elements.

[0091] Set the loss function and optimizer. The loss function is the cross-entropy loss function, and the optimizer is the Adam algorithm.

[0092] The model is trained using a training set, with 10–20 samples used in each iteration. After each training iteration, the training loss is calculated and tested on a test set. If the training loss converges and the average test accuracy exceeds a set threshold of 95%, the model training is stopped, and the trained FNN-based classification and recognition model is output.

[0093] S103: Based on the image labels and the center wavelength set by the spectrometer, use a scatter plot to draw the PC rotation angle combination of all images with the classification label as mode-locked. The data corresponding to different center wavelengths are marked with different colors. Obtain the PC angle intervals corresponding to each center wavelength that show the mode-locked state as the mode-locked intervals.

[0094] S104: Specify the center wavelength, select the corresponding mode-locked interval according to the specified center wavelength, and control the servo motor to traverse the PC rotation angle combination in an orderly manner according to a fixed step size through the program. When the target image appears, stop the PC rotation to achieve fiber mode-locked tuning.

[0095] like Figure 5 As shown, the process of controlling the servo motor to sequentially traverse the PC rotation angle combinations in a fixed step size is illustrated:

[0096] Initialize the servo angle and restore it to the vertical position;

[0097] Determine the required center wavelength and select the corresponding mode-locked interval as the input to the traversal control algorithm;

[0098] The servo motor is controlled by a traversal control algorithm to traverse the PC angle combination with a change step size of 1° to 1.5°. Each time the PC angle changes, the image output by the spectrometer is detected in real time.

[0099] If the output image is detected to be similar to the target image, the program terminates.

[0100] If the output image is found to be dissimilar to the target image, the detection continues until the program ends.

[0101] Specifically, the present invention will be further illustrated below through embodiments:

[0102] To verify the effectiveness of the method proposed in Example 1, this example successfully implemented tunable dual-wavelength intelligent mode-locking, reproducing the single soliton mode-locked spectra in two bands. For example... Figure 6 As shown, when the desired center wavelength is specified to be around 1530nm, the experimental results are as follows. Figure 6 As shown by the gray curve, the center wavelength is 1530.9 nm; when the desired center wavelength is specified to be around 1560 nm, the experimental results are as follows. Figure 6 As shown by the black curve, the center wavelength is 1562.2nm.

[0103] This invention focuses on the differences in spectral images between mode-locked and mode-free states. Based on a fully connected neural network with linear transformation, it constructs a recognition and classification model that accurately identifies the mode-locked state within a fiber laser cavity. In this process, the RGB spectral image undergoes HSV color space conversion, and a filtering algorithm is used to extract the spectral curve image, significantly reducing background and noise interference in the image to be identified and improving the training and computation speed of the model. The fastest training time for this model is 49.65 seconds, a significant reduction compared to traditional simulation methods that often take several hours.

[0104] Simultaneously, since this method defines the mode-locking intervals for each wavelength, different soliton mode-locking images can be adjusted by traversing the control program within these intervals. Therefore, the center wavelengths of all soliton mode-locking can be calculated, yielding the center wavelength range for each mode-locking band. For example... Figure 7 As shown, the center wavelength boundaries for mode-locking in the 1530nm band are 1529.4nm and 1531.9nm; Figure 8 As shown, the center wavelength boundaries of mode-locking in the 1560nm band are 1560.0nm and 1563.3nm. In this embodiment, within both ranges, the mode-locking state with different center wavelengths can be tuned by fine-tuning the PC angle through an ergonomic control algorithm.

[0105] Example 2: Second aspect, such as Figure 9 As shown, in order to achieve the above objectives, this invention discloses a system for tuning fiber mode-locked lasers based on image recognition algorithms, comprising:

[0106] Image processing module 11 is used to receive spectrometer images, preprocess the spectrometer images to obtain preprocessed spectrometer images, generate image datasets from the preprocessed spectrometer images, and perform labeling operations on the image datasets to obtain labeled image datasets. The spectrometer images are acquired based on the center wavelength set by the spectrometer.

[0107] The model training module 12 is used to input the labeled image dataset into the pre-established FNN-based classification and recognition model for training, obtain the trained FNN-based classification and recognition model, acquire the spectrometer image to be identified, input the spectrometer image to be identified into the trained FNN-based classification and recognition model, and output the image label.

[0108] The interval determination module 13 is used to draw a scatter plot of all PC rotation angle combinations of images with the classification label as mode-locked based on the image label and the center wavelength set by the spectrometer. The data corresponding to different center wavelengths are marked with different colors, and the PC angle intervals corresponding to each center wavelength that show the mode-locked state are obtained as the mode-locked intervals.

[0109] The mode-locked tuning module 14 specifies the center wavelength, selects the corresponding mode-locked interval according to the specified center wavelength, and controls the servo motor to traverse the PC rotation angle combination in an orderly manner according to a fixed step size through the program. When the target image appears, the PC rotation stops, thereby realizing the mode-locked tuning of the fiber.

[0110] Based on the same inventive concept, this invention also provides a computer device, comprising: one or more processors, and a memory for storing one or more computer programs; the programs include program instructions, and the processor executes the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, used to implement one or more instructions, specifically for loading and executing one or more instructions stored in a computer storage medium to implement the above-described method.

[0111] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, performs the above-described method. This storage medium can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0112] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0113] The foregoing has shown and described the basic principles, main features, and advantages of this disclosure. Those skilled in the art should understand that this disclosure is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this disclosure. Various changes and modifications can be made to this disclosure without departing from its spirit and scope, and all such changes and modifications fall within the scope of this disclosure as claimed.

Claims

1. A method for tuning fiber mode-locked laser based on image recognition algorithm, characterized in that, The method includes the following steps: Receive spectrometer images, preprocess the spectrometer images to obtain preprocessed spectrometer images, generate image datasets from the preprocessed spectrometer images, and perform labeling operations on the image datasets to obtain labeled image datasets, wherein the spectrometer images are acquired based on the center wavelength set by the spectrometer. The labeled image dataset is input into a pre-established FNN-based classification and recognition model for training, resulting in a trained FNN-based classification and recognition model. The spectrometer image to be identified is then obtained, input into the trained FNN-based classification and recognition model, and the image label is output. Based on the image labels and the center wavelength set by the spectrometer, a scatter plot was used to draw the PC rotation angle combination of all images with the classification label as mode-locked. Data corresponding to different center wavelengths were marked with different colors, and the PC angle intervals in which the mode-locked state appeared for each center wavelength were obtained as the mode-locked intervals. A center wavelength is specified, and the corresponding mode-locking interval is selected according to the specified center wavelength. The servo motor is controlled by the program to traverse the PC rotation angle combination in an orderly manner according to a fixed step size. When the target image appears, the PC rotation stops, thus realizing the tuning of fiber mode-locking.

2. The method for tuning fiber mode-locked laser based on image recognition algorithm according to claim 1, characterized in that, The process of acquiring the spectrometer images is as follows: The center wavelength of the spectrometer is set to be near the center wavelength of the two types of solitons generated by the laser cavity. The servo motor is controlled by the program to rotate the PC in an orderly manner with a fixed step size, and a spectrometer image is acquired once for each rotation.

3. The method for tuning fiber mode-locked laser based on image recognition algorithm according to claim 2, characterized in that, The process of preprocessing the spectrometer images: The input image is converted from its RGB color space spectrum to the HSV color space using the following formula: In the formula, R, G, and B are the values ​​of an image in the red, green, and blue color components, respectively, and H, S, and V are the values ​​of an image in the hue, saturation, and lightness components, respectively. By setting upper and lower thresholds for the H, S, and V parameters, and using mask extraction, the image containing only the spectral curve is separated from the converted image. By setting a rectangular convolution kernel and performing morphological operations such as opening, the separated image is filtered to remove noise. Use Crop to crop the center of the image to obtain a square-sized image of the central portion; The obtained image is converted into a grayscale image for output.

4. The method for tuning fiber mode-locked laser based on image recognition algorithm according to claim 1, characterized in that, The FNN-based classification and recognition model is a fully connected neural network based on linear image transformation. The fully connected neural network based on linear image transformation includes: an input layer, a linear layer, four hidden layers, a softmax layer, and an output layer.

5. The method for tuning fiber mode-locked laser based on image recognition algorithm according to claim 1, characterized in that, The process of inputting the labeled image dataset into a pre-established FNN-based classification and recognition model for training: The labeled image dataset is divided into a training set and a test set; The fully connected neural network based on image linear transformation receives image data represented by a two-dimensional matrix in its input layer, and the linear layer transforms the two-dimensional matrix into a one-dimensional array which is then input into the hidden layer. Each neuron in each layer is fully connected, and the activation function is the ReLU function. The mapping relationship of the functions is as follows: in, x The input to each neuron, f(x) The output after passing through the neuron; The final activation function is the tanh function, and the mapping relationship of the functions is as follows: in, x The input to each neuron, f(x) The output after passing through the neuron; sinh(x) It is a hyperbolic sine function. cosh(x) It is a hyperbolic cosine function. tanh(x) It is the hyperbolic tangent function; The output layer uses the softmax function to convert the output of the previous neural network layer into probability values ​​between 0 and 1. P i The formula is: in, It is a vector, representing the output of the previous layer of the neural network. and It is one of the elements; the z vector contains two elements. Set the loss function and optimizer. The loss function is the cross-entropy loss function, and the optimizer is the Adam algorithm. The model is trained using the training set. After each training iteration, the training loss is calculated and tested on the test set. If the training loss converges and the average test accuracy exceeds a set threshold, the model training is stopped, and the trained FNN-based classification and recognition model is output.

6. The method for tuning fiber mode-locked laser based on image recognition algorithm according to claim 1, characterized in that, The process of controlling the servo motor to sequentially traverse the PC rotation angle combination according to a fixed step size: Initialize the servo angle and restore it to the vertical position; Determine the required center wavelength and select the corresponding mode-locked interval as the input to the traversal control algorithm; The servo motor is controlled by a traversal control algorithm to traverse the PC angle combination with a change step size of 1°~1.5°. Each time the PC angle changes, the image output by the spectrometer is detected in real time. If the output image is detected to be similar to the target image, the program terminates. If the output image is found to be dissimilar to the target image, the detection continues until the program ends.

7. A system for tuning fiber mode-locked laser based on image recognition algorithm, characterized in that, include: The image processing module is used to receive spectrometer images, preprocess the spectrometer images to obtain preprocessed spectrometer images, generate image datasets from the preprocessed spectrometer images, and perform labeling operations on the image datasets to obtain labeled image datasets. The spectrometer images are acquired based on the center wavelength set by the spectrometer. The model training module is used to input the labeled image dataset into a pre-established FNN-based classification and recognition model for training, obtain the trained FNN-based classification and recognition model, acquire the spectrometer image to be identified, input the spectrometer image to be identified into the trained FNN-based classification and recognition model, and output the image label. The interval determination module is used to plot the PC rotation angle combination of all images with the classification label as mode-locked based on the image label and the center wavelength set by the spectrometer. The data corresponding to different center wavelengths are marked with different colors, and the PC angle intervals corresponding to each center wavelength that show the mode-locked state are obtained as the mode-locked intervals. The mode-locked tuning module allows you to specify a center wavelength, select the corresponding mode-locked interval based on the specified center wavelength, and control the servo motor to traverse the PC rotation angle combination in an orderly manner with a fixed step size. When the target image appears, the PC rotation stops, thus achieving fiber mode-locked tuning.

8. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, The memory stores a computer program that can run on a processor. When the processor loads and executes the computer program, it employs a method for tuning fiber mode-locked laser based on an image recognition algorithm, as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is loaded and executed by the processor, it employs a method for tuning fiber mode-locked laser based on an image recognition algorithm, as described in any one of claims 1 to 6.

Images