Thin-film lithium niobate chip fiber-coupled machine vision alignment control method

Abstract

This disclosure provides a machine vision alignment control method for fiber optic coupling of a thin-film lithium niobate chip. The method includes acquiring a high-resolution image of the waveguide end face of the thin-film lithium niobate chip and extracting a set of multi-structure visual representations from it; determining the effective mode field region by combining the emitted light intensity distribution with the geometric constraints provided by the sidewall boundaries; calculating the energy weighting center based on the light intensity distribution within the effective mode field region and mapping this energy weighting center to the assembly platform coordinate system to obtain an optical coupling reference center for alignment; adjusting the position and orientation of the optical fiber using feedback control with the optical coupling reference center as the target, so that the fiber core center approaches and aligns with the optical coupling reference center; evaluating and locking a stable fiber pose after alignment, and outputting the final coupling pose and performance indicators. This method can overcome the decoupling problem between the visual reference and the real optical reference, achieving low-loss, high-stability coupling between the thin-film lithium niobate chip and the optical fiber.

Description

Technical Field

[0001] This disclosure relates to the fields of optoelectronic device manufacturing and precision optical assembly technology, and in particular to a machine vision alignment control method for fiber-optic coupling of thin-film lithium niobate chips. Background Technology

[0002] Due to their excellent electro-optical properties, thin-film lithium niobate chips have become core devices in fields such as high-speed optical communication and integrated photonics. To achieve efficient optical signal transmission between the chip and external optical fibers, precise sub-micron alignment of the fiber core and the chip waveguide end face is essential. Currently, mainstream automated alignment schemes rely on machine vision technology, which uses microscopic imaging to acquire images of the chip waveguide end face and identifies the geometric center of the waveguide structure in the image as the reference position for optical coupling, thereby controlling the fiber to align to that position.

[0003] However, thin-film lithium niobate chips are generally fabricated using deep etching processes, resulting in waveguides with high refractive index contrast. Their end faces are not ideally smooth planes, but rather exhibit complex morphologies such as significant etching steps, sidewall tilt angles, and surface roughness. In such practical structures, the propagation and distribution of the light field are strongly modulated by abrupt changes in refractive index and complex boundary structures. This leads to a systematic deviation between the actual output mode field center—the optimal location for optical coupling—and the geometric end face center identified by simple image processing. This deviation is not a fixed constant but dynamically changes with the specific width of the waveguide, the etching depth, and even the incident conditions of the optical signal. Therefore, even if existing methods can drive the fiber to be visually precisely aligned with the geometric center and perform local optimization under power monitoring, it may still ultimately remain in a suboptimal mode field overlap state due to reference misalignment. This introduces a subtle but performance-impacting implicit insertion loss, limiting the device's coupling efficiency and performance consistency and yield.

[0004] Therefore, there is an urgent need for an alignment control method that can overcome the decoupling problem between the visual reference and the real optical reference, so as to achieve truly low-loss and high-stability coupling between thin-film lithium niobate chips and optical fibers. Summary of the Invention

[0005] In view of this, in order to solve the problems caused by the prior art, this application provides a thin-film lithium niobate chip fiber-coupled machine vision alignment control method.

[0006] In a first aspect, this disclosure provides a method for aligning and controlling a thin-film lithium niobate chip with fiber-optic coupling in machine vision, the method comprising: S1. Acquire high-resolution images of the waveguide end face of the thin-film lithium niobate chip, extract the etched step contour, sidewall boundary, and normalized light intensity distribution from them to form a set of multi-structure visual representations. S2. Based on the multi-structure visual representation set, and combining the emitted light intensity distribution with the geometric constraints provided by the sidewall boundaries, determine the effective area of ​​the mode field. S3. Based on the light intensity distribution within the effective area of ​​the mode field, calculate its energy weighting center and map the energy weighting center to the assembly platform coordinate system to obtain the optical coupling reference center for alignment. S4. Using the optical coupling reference center as the target, adjust the position and attitude of the optical fiber through feedback control so that the center of the fiber core approaches and aligns with the optical coupling reference center. S5. After alignment is completed, evaluate and lock the stable fiber pose, and output the final coupling pose and performance indicators.

[0007] Optionally, S1 includes: The grayscale image of the waveguide end face is acquired by a microscopic imaging system, the pixel coordinate system of the image is used as a unified spatial reference frame, and the conversion factor from pixel to physical length is calculated. In the pixel coordinate system, the etched step contour is extracted based on gradient operation and threshold segmentation, and the sidewall boundary of the waveguide is extracted based on the step contour through local constraint search. Within the area defined by the sidewall boundary, the pixel values ​​of the grayscale image are normalized to obtain the normalized light intensity distribution, and its energy weighting center is calculated. The etched step contour, sidewall boundary, normalized light intensity distribution and its energy weighting center are integrated in the unified coordinate system to form a multi-structure visual representation set.

[0008] Optionally, S2 includes: Based on the waveguide sidewall boundary in the multi-structure visual representation set, an effective analysis window for the end face is delineated in the image. Within the window, based on the normalized light intensity value of each pixel, the energy coverage ratio threshold method is used to filter out the candidate set of light-emitting core areas whose cumulative energy reaches a preset ratio; Based on the area covered by the candidate set of the light-emitting core region, the minimum distance from each pixel in the window to the sidewall boundary and the etched step contour is calculated, and a structural weight function is constructed based on the distance. The normalized emitted light intensity distribution within the window is corrected using the structure weight function to obtain the structure-related corrected emitted light distribution. The final effective mode field region and its boundary are then determined using the energy coverage ratio threshold method.

[0009] Optionally, S3 includes: The effective area of ​​the mode field and its corresponding normalized light intensity weight distribution are transformed from image pixel coordinates to end face physical coordinates; In the physical coordinates of the end face, its energy weighting center is calculated based on the transformed weight distribution and used as the equivalent optical coupling reference center; In the physical coordinates of the end face, the equivalent uncertainty radius characterizing the spatial discreteness of the optical field is calculated based on the transformed weight distribution; Based on the equivalent optical coupling reference center and the equivalent uncertainty radius, the end face spatial alignment with the target is formed; The end face space alignment target is mapped to the assembly platform coordinate system to obtain the optical coupling reference center and assembly tolerance radius in the platform coordinate system used for direct alignment control.

[0010] Optionally, the calculation of the equivalent uncertainty radius characterizing the spatial discreteness of the light field based on the transformed weight distribution includes: Calculate the lateral weighted variance and longitudinal weighted variance of the weight distribution in the physical coordinates of the end face relative to the equivalent optical coupling reference center; The equivalent uncertainty radius is synthesized from the horizontal weighted variance and the vertical weighted variance.

[0011] Optionally, S4 includes: Real-time acquisition of the current position coordinates of the fiber optic core center in the assembly platform coordinate system; Calculate the positional deviation between the current position and the optical coupling reference center in the platform coordinate system; Based on the magnitude of the positional deviation, a segmented and step-adaptive strategy is adopted to control the actuator to drive the optical fiber to translate and approach, so that the center of the optical fiber core enters the area defined by the optical coupling reference center as the center and the assembly tolerance radius as the radius.

[0012] Optionally, after the process of moving the fiber core center into the region defined by the optical coupling reference center and the assembly tolerance as the radius, the method further includes: Calculate the total light intensity in the left and right regions bounded by the horizontal coordinate of the optical coupling reference center, and the upper and lower regions bounded by the vertical coordinate, and calculate the left-right energy difference and the upper-lower energy difference. Divide the energy difference by the total emitted light intensity within the entire effective area of ​​the mode field to obtain the normalized left-right asymmetry and up-down asymmetry. The yaw angle adjustment is calculated based on the left-right asymmetry, the pitch angle adjustment is calculated based on the up-down asymmetry, and attitude fine-tuning is performed accordingly. When the adjusted translational residual deviation is less than the assembly tolerance radius, and the absolute values ​​of the adjusted left-right and up-down optical field distribution asymmetry are both lower than the preset threshold, the fiber pose alignment is determined to be complete.

[0013] Optionally, S5 includes: After the alignment is determined, the instantaneous distance fluctuation between the fiber core center and the optical coupling reference center is monitored over multiple consecutive sampling periods. Calculate the root mean square value of the consecutive instantaneous distances as a stability evaluation metric; When the stability evaluation value is less than or equal to a preset stability threshold, it is determined that a stable coincidence state has been reached. For fiber pose when a stable coincidence state is reached, a locking operation is performed. The fiber core center coordinates and attitude angles continuously sampled during the stable state are smoothed and averaged, and then solidified into the final coupling alignment parameter set. Under the pose defined by the final coupling alignment parameter set, the input power of the optical signal and the output power after optical fiber coupling are measured, and the average insertion loss is calculated. The output contains the final result including the final set of coupling alignment parameters and the average insertion loss.

[0014] In a second aspect, this disclosure provides an electronic device including a memory and at least one processor, the memory storing a computer program, and the processor executing the computer program to implement the method of the first aspect described above.

[0015] Thirdly, this disclosure provides a computer storage medium storing a computer program that, when executed, implements the method described in the first aspect.

[0016] The beneficial effects of this disclosure are that, compared with the prior art, this disclosure has the following advantages: 1) By constructing a set of multi-structure visual representations in a unified coordinate system that integrates the contour of the etched steps on the waveguide end face, the sidewall boundary, and the normalized light intensity distribution, the limitations of traditional methods that rely solely on a single geometric feature for alignment reference identification are overcome. This provides a complete data foundation for subsequent analysis that simultaneously includes the fine geometric shape of the end face and the real light field distribution information, enabling the identification of the alignment reference to fully reflect the complex structural influence brought about by the deep etching process, thus laying a data foundation for accurately locating the effective optical area.

[0017] 2) A method is proposed to determine the effective operating area of ​​the mode field based on the synergistic effect of sidewall geometric constraints and light intensity distribution. By introducing structural correlation weights to correct the candidate light-emitting core area, the interference of the real optical operating area and the interference of light intensity signals such as reflection and scattering caused by etching morphology on the positioning of the real optical mode field center is effectively filtered out. This solves the systematic deviation problem between the real light-emitting mode field center and the simple image geometric center caused by structural modulation, and ensures the accuracy of the optical reference in principle.

[0018] 3) By calculating the energy weighting center of the effective region of the mode field in physical coordinates and synthesizing the equivalent uncertainty radius characterizing the spatial dispersion of the optical field, an optical coupling reference center with both clear positional information and a reasonable tolerance range is generated. This reference center is then accurately mapped to the assembly platform coordinate system through a pre-calibrated coordinate transformation relationship, providing a direct, stable, and physically clear spatial reference (including center coordinates and tolerance radius) for motion control. This replaces the traditionally used, easily misaligned geometric center reference, thus avoiding implicit insertion loss and coupling performance fluctuations caused by reference inaccuracy at the source.

[0019] 4) Based on the aforementioned optical coupling reference center, feedback-based alignment control is implemented. A strategy combining segmented translational approximation with attitude fine-tuning based on the symmetry of the emitted light distribution is employed to achieve phased, collaborative optimization of the fiber pose in translational and angular degrees of freedom. This method not only enables the fiber core to rapidly and stably approach the optimal optical position but also actively suppresses coupling degradation caused by angular mismatch. After alignment, by continuously monitoring pose fluctuations and quantitatively evaluating stability, the optimal coupling pose is ultimately locked, thus achieving true sub-micron level precision alignment and significantly improving final coupling efficiency, consistency, and long-term stability. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0021] Figure 1 A flowchart of the thin-film lithium niobate chip fiber-coupled machine vision alignment control method provided in an embodiment of this disclosure is shown. Figure 2 A flowchart of fiber optic pose feedback control alignment provided in an embodiment of this disclosure is shown.

[0022] The accompanying drawings have illustrated specific embodiments of this disclosure, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this disclosure to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0023] The present disclosure will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present disclosure and should not be used to limit the scope of protection of the present disclosure.

[0024] The components of the embodiments of the invention described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0025] In the following, the terms “comprising,” “having,” and their cognates, which may be used in various embodiments of the invention, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as excluding, firstly, the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more features, numbers, steps, operations, elements, components, or combinations thereof.

[0026] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0027] Figure 1 A flowchart of the thin-film lithium niobate chip fiber-coupled machine vision alignment control method provided in this disclosure embodiment is shown below. Figure 1 As shown, the process may include the following steps: S1: Acquire high-resolution images of the waveguide end face of the thin-film lithium niobate chip, extract the etched step contour, sidewall boundary, and normalized light intensity distribution from them to form a set of multi-structure visual representations.

[0028] High-resolution imaging is used to acquire images of the waveguide end face, and the etched step contour, sidewall boundary, and effective light intensity distribution are extracted in the same coordinate system to form a unified visual representation set, providing a data foundation for subsequent analysis. This is achieved through the following sub-steps.

[0029] S1.1: Perform high-resolution imaging of the waveguide end face and establish a unified coordinate system and pixel physical conversion relationship.

[0030] The pre-mounted and pre-focused thin-film lithium niobate chip waveguide end face is placed under a microscopic imaging system. This system is equipped with a high numerical objective lens and a scientific-grade camera to ensure image sharpness and resolution. The imaging magnification M is set, typically ranging from 10 to 50 times to accommodate waveguide structures of different sizes. Simultaneously, the pixel size p is set according to the camera characteristics, generally between 1.5 and 5 micrometers. Exposure time... The intensity of light emitted from the end face is then adjusted, typically within the range of 0.5 milliseconds to 20 milliseconds, to obtain an image with a good signal-to-noise ratio.

[0031] The imaging system is activated to acquire a grayscale image of the end face, denoted as... ; Here, x and y are the horizontal and vertical pixel indices of the image, respectively, and their maximum values ​​are determined by the image width W and height H. After acquisition, the coordinate system definition of the image is immediately fixed, serving as the unified spatial reference framework for all subsequent feature extraction and analysis.

[0032] To convert pixel scale to true physical scale, a conversion factor 's' from pixel to physical length needs to be calculated, with units of micrometers per pixel. This factor is obtained by the ratio of the camera pixel size 'p' to the microscope imaging magnification 'M', and the calculation formula is as follows: Where p is the camera pixel size and M is the magnification. The calculated s value will be used to subsequently convert all pixel-based feature coordinates and dimensions into physical quantities in micrometers, thereby ensuring that the visual representation is accurately correlated with the physical coordinate system of the final assembly platform. This step transforms the seen end face into an image basis with well-defined physical metrics, suitable for engineering analysis.

[0033] S1.2: Extract the contour of the end face etching step by gradient enhancement and threshold segmentation.

[0034] Based on the high-resolution end-face grayscale image obtained above The next step is to extract the step contours formed by the deep etching process. Steps appear as drastic changes in grayscale values ​​on the image, so gradient operations are used to enhance these edge features.

[0035] In practice, the edge features in the image are first enhanced through gradient calculation, and the calculation is performed for each pixel. gradient magnitude at .

[0036] The gradient magnitude reflects the degree of grayscale change at that point. It can be approximated by calculating the sum of the squares of the grayscale differences between the pixel and its neighboring pixels in the horizontal and vertical directions, and then taking the square root.

[0037] After calculating the gradient magnitude of the entire image, a gradient threshold is set. To identify and extract step boundaries.

[0038] threshold The value ranges from 10 to 60 grayscale difference units. In practice, it can be determined by analyzing the average gradient magnitude of non-step flat areas on the same chip and adding an engineering margin.

[0039] All pixels with gradient magnitudes greater than or equal to the threshold are identified as belonging to the step boundary, and the set of these points constitutes the etching step contour set. Its mathematical expression is: ; To evaluate the continuity of the extracted step contour, the total number of pixels in the set can be counted, denoted as the step boundary point count. The calculation method is to calculate all elements in the image that satisfy... The number of pixels is counted.

[0040] This step outputs a set of contours. The location and orientation of all etched steps on the end face were clearly marked, providing the first layer of key geometric constraint information for subsequent analysis of the waveguide structure.

[0041] S1.3: Local search and extraction of waveguide sidewall boundary lines based on step profile.

[0042] After obtaining the step profile, it is necessary to further extract the boundary lines of the waveguide sidewalls. Since the end face may have sidewall tilt angles or etch roughness, the sidewall boundaries are a more stable reflection of the waveguide's true physical constraints than the step profile. To avoid misidentifying noise or other non-waveguide structures in the image as sidewalls, this step employs a local constraint search strategy.

[0043] Specifically, the set of step contours obtained in the previous step Each boundary point in Starting from a reference point, a search is performed within a local neighborhood along its normal direction. The goal is to find pixels with continuous grayscale changes and stable gradient magnitudes, as these points are more likely to belong to the true sidewall surface. This operation is performed on both the left and right walls of the waveguide. The searched pixels are then aggregated to obtain the boundary point sets for the left and right walls, which are ultimately merged into a complete set of sidewall boundary lines. .

[0044] To ensure the engineering usability of the extracted sidewall boundary lines, i.e., that their orientation is generally stable and consistent rather than chaotic, a sidewall orientation consistency index can be calculated. A quantitative evaluation is conducted. The calculation method for this index is as follows: first, calculate the local azimuth angle at each sidewall boundary point; then, calculate the absolute deviation of all these local azimuth angles from their average value; finally, calculate the average of these absolute deviations. The result is the sidewall orientation consistency index. ; The smaller the value, the more consistent the sidewall orientation. In engineering experience, when... At this point, the extracted sidewall boundaries can be considered to have good stability and can be used in subsequent steps. This step outputs a set of sidewall boundary lines. This provides an accurate anchor point for subsequently associating and constraining the light distribution with specific waveguide physical structures.

[0045] S1.4: Construct a normalized light intensity distribution within the sidewall constraint region and summarize all features to form a visual representation set.

[0046] Under the same image coordinate system, the effective light intensity distribution of the waveguide end face is constructed, and all extracted structural and optical field information is summarized.

[0047] Based on the set of sidewall boundary lines The physical region of the waveguide sidewall, as characterized, is identified as the potential light emission range of the waveguide core. Here, by... Within the defined area, the original grayscale image The pixel values ​​are normalized to obtain a normalized emitted light intensity distribution. The normalization formula is as follows: ; The denominator represents the summation of all pixels within the physical area enclosed by the sidewall boundary.

[0048] After normalization, The value of is between 0 and 1, and it applies to all points within that region. The sum of the values ​​is 1, which gives it the physical meaning of probability weights, representing the relative contribution of each pixel position to the total emitted light energy.

[0049] Next, based on this intensity distribution, the coordinates of the energy-weighted center of the light region are calculated. Center coordinates Obtained through intensity-weighted averaging: ; ; The calculated center coordinates are in pixels. If physical coordinates are needed, they can be multiplied by the conversion factor s obtained in step S1.1.

[0050] The original end face grayscale image obtained in this step Etched step outline set Set of sidewall boundary lines Normalize the light intensity distribution and the coordinates of the light output center Packed together in a unified image coordinate system, a visual representation set of multiple structural features of the waveguide end face is formed. This dataset comprehensively integrates the geometric morphology information and light field distribution information of the end face. The former clarifies the spatial location and boundaries of the structure, while the latter reveals the effective emission location and relative intensity of the light signal, thus laying a solid data foundation for accurately identifying the effective area of ​​the real optical mode field in subsequent steps.

[0051] In the technical solution of this disclosure, by using high-resolution imaging and establishing a unified coordinate system, combined with gradient enhancement and boundary extraction, bilateral constraint extraction of sidewall boundary lines, and construction of intensity distribution in the effective light-emitting region, a visual representation set that comprehensively integrates end-face geometry and light field distribution information is formed. This set not only accurately calibrates the key geometric constraints of the etching steps and sidewall boundaries, but also quantitatively characterizes the normalized light intensity distribution and its energy weighting center, thus providing a reliable data foundation with both spatial location information and optical weighting information for subsequent steps, fundamentally avoiding the limitations of relying solely on a single geometric feature for analysis.

[0052] S2: Based on the multi-structure visual representation set, combined with the light intensity distribution and the geometric constraints provided by the sidewall boundaries, the effective area of ​​the mode field is determined.

[0053] Based on the visual representation set obtained in step S1, and combining the emitted light intensity distribution with the geometric constraints of the end face, the effective area of ​​the real light field on the end face is identified and defined, and an optical reference area is established. This is specifically achieved through the following sub-steps.

[0054] S2.1: Generate an effective analysis window for the end face based on the sidewall boundary constraints and normalize the light distribution within the window.

[0055] Using the set of sidewall boundary lines obtained in step S1 The provided spatial constraint information is problematic. Due to the etching steps and surface roughness on the end face, directly analyzing the light distribution across the entire image range could easily misjudge stray reflected light or localized highlights in the etching texture as effective light-emitting areas, thus introducing interference. Therefore, it is necessary to first define a physical analysis range that only includes the possible areas of actual wavelet-derived light.

[0056] In practice, based on the sidewall boundary, a closed region enclosed by the sidewall structure is formed by appropriate inward extension and connection, and this region is defined as the effective analysis window of the end face. The window is geometrically strictly constrained by the physical sidewalls of the waveguide, thus imposing a hard constraint on subsequent analysis in space.

[0057] Next, in this window Internally, based on the light intensity distribution data obtained in step S1 Normalization is performed to obtain the normalized output light distribution within the window. Its calculation method is to convert the original distribution... Except for its window The sum of all pixel values ​​within the range is expressed by the following formula: ; After this treatment In the window The value range within the range is 0 to 1, and the sum of these values ​​is 1. This ensures that all subsequent calculations based on energy weights only reflect the characteristics of the real, effective light-emitting region constrained by the waveguide sidewalls.

[0058] S2.2: The energy coverage ratio threshold method is used to extract the candidate set of strongly correlated light-emitting core regions.

[0059] Within the limited effective analysis window, it is necessary to further focus on the core region that bears the majority of the emitted light energy. The core engineering meaning of the effective mode field region is that it should bear the majority of the transmitted light energy on the end face. Therefore, this sub-step uses the energy coverage ratio threshold method to define the core region, rather than a simple fixed grayscale threshold, to avoid being misled by local bright noise points on the end face.

[0060] In practice, first open the window The light intensity value of each pixel within is calculated according to its corresponding normalization. Sort the intensity values ​​from largest to smallest. Then, starting with the maximum value, accumulate these intensity values ​​until the total reaches a preset target ratio. . The value ranges from 0.70 to 0.90, and 0.80 is commonly used in engineering practice. The smallest set of pixels that meets this condition is defined as the candidate set of the light-emitting core region. In other words, this set contains the fewest pixels, but its cumulative normalized light intensity is not less than [amount missing]. The set of pixels. Simultaneously, the total number of pixels contained in this core area can be counted, denoted as . This step is used to initially determine whether the region is spatially too fragmented and dispersed or excessively expanded. This step yields a candidate set of core regions. It screened out the spatial regions most strongly associated with real optical light output from the perspective of energy contribution, providing initial, physically meaningful candidate targets for subsequent correction based on structural morphology.

[0061] S2.3: Establish structural correlation constraints between light distribution and end face steps and sidewalls, and refine the candidate set.

[0062] Traditional alignment methods typically consider only the emitted light intensity distribution, neglecting the modulation effect of the end-face physical structure on the spatial morphology of the light field. Therefore, candidate regions defined solely based on intensity may be located in structurally unsuitable positions, such as being too close to sidewalls or steps, leading to mode field distortion or scattering loss, thus affecting the final coupling efficiency. This sub-step aims to explicitly introduce this spatial constraint into the region definition process.

[0063] In practice, it is necessary to calculate each pixel within the effective analysis window. The minimum Euclidean distance to the boundaries of two types of critical structures. One is the distance to the set of boundaries representing the sidewalls. minimum distance Secondly, it is to obtain the set of step outlines. minimum distance .

[0064] Based on these two distances, a structural weight function is constructed. The design intent is to make the candidate region more likely to fall in a reasonable position that is neither too close to the sidewall to avoid mode distortion, nor too close to the step to avoid strong scattering loss. The weighting function is defined as follows: ; Here, a and b are distance smoothing constants, typically ranging from 3 to 15 pixels, used to prevent the weight from becoming zero when the distance is too small. The value ranges from 0 to 1. The larger the value, the more favorable the spatial relationship between the position and the main structure of the end face.

[0065] Subsequently, the structural weights are multiplied by the normalized emission distribution to obtain the emission distribution corrected for structural correlation. .

[0066] Next, the same energy coverage ratio threshold as in sub-step S2.2 is used. In the corrected distribution The set of pixels whose cumulative energy reaches the threshold is reselected to obtain the candidate set after structure-related correction. .

[0067] To quantify the effectiveness of this correction, a structural relevance index is defined. Its value is in the corrected core region within, correct the light distribution Energy and original effective light output distribution The ratio of their energies.

[0068] The closer the value is to 1, the more it indicates that the correction process has not significantly weakened the effective light output energy, meaning that the light output mainly falls in the area with a reasonable structural relationship; the lower the value, the more it suggests that the light output energy is mainly distributed in the area with an unfavorable structure, which needs to be paid attention to in the process or assembly.

[0069] S2.4: Effective area of ​​the output mode field, its boundary set, and energy coverage index.

[0070] Candidate set corrected for structural correlation The main body identified as the effective area of ​​the model field is denoted as... To ensure the region is easily interpretable and manageable in engineering, connectivity analysis or morphological adjustments are typically performed to make it more accessible. Ultimately, it manifests as one or more connected regions.

[0071] It is necessary to extract the set of pixels with defined boundaries from this region. The boundary point is the region. At least one of its four neighboring pixels does not belong to The pixels.

[0072] Meanwhile, to confirm that this final area indeed bears the majority of the light energy, its energy coverage index is calculated. That is, the normalized light intensity in this region The sum: ; The value ranges from 0 to 1, and it is recommended in engineering to have a value of no less than 0.70 to ensure that the defined area is sufficiently representative.

[0073] This step yields the effective area of ​​the end face mold field. Its boundary set and energy coverage indicators Together, these constitute an effective optical reference region, defined by the light output distribution and end-face structural features, distinct from a simple geometric center. This region provides a direct and physically consistent spatial input for the next step of calculating the optical coupling reference center.

[0074] In the technical solution of this disclosure embodiment, based on the sidewall constraints and light emission distribution in the visual representation set, an effective analysis window of the end face is generated, a strongly correlated light emission core area is extracted, and further, the structural correlation constraints between the light emission distribution and the end face etching steps and sidewalls are established. Finally, an effective mode field region and its boundary defined by the light field energy distribution and the end face physical structure are determined. This overcomes the defect of the traditional scheme that only considers light intensity and ignores structural modulation, making the defined optical effective region physically self-consistent, eliminating interfering light intensity signals located in unfavorable structural regions, and laying a regional foundation for obtaining a real optical coupling reference.

[0075] S3: Based on the light intensity distribution within the effective area of ​​the mode field, calculate its energy weighting center and map the energy weighting center to the assembly platform coordinate system to obtain the optical coupling reference center for alignment.

[0076] Based on the effective area determined in step S2, the energy weighting center of its light intensity distribution is calculated as the optical coupling reference center, and this center is mapped from the end face image coordinate system to the assembly platform coordinate system to form a spatial reference that can be directly used for alignment. This is specifically achieved through the following sub-steps.

[0077] S3.1: Transform the effective area of ​​the model field to the physical scale coordinate field and associate the weight distribution.

[0078] The effective area of ​​the mode field output in step S2 The image space is transformed from pixel to micrometer-scale coordinate space with a defined physical scale to align with the physical dimensions of the assembly platform. This transformation relies on the pixel-to-physical-length conversion factor s obtained in step S1.1, which is in micrometers per pixel.

[0079] For the region Each pixel within By multiplying by a conversion factor s, its coordinates are converted into physical coordinates on the end face. The conversion formula is: , Where u and v are in micrometers. All transformed coordinate points constitute the physical coordinate point set. .

[0080] Meanwhile, the region The normalized light intensity weight corresponding to each pixel. They are inherited and associated with their corresponding physical coordinate points, forming a weight distribution under physical coordinates. This weight distribution characterizes the relative contribution of each point in space to the overall emitted light energy at the micrometer scale, and satisfies the condition in the set... The sum of all values ​​is 1. The purpose of establishing a physical scale coordinate field is to provide a unified measurement basis that can be directly used for engineering assembly for all subsequent calculations, and to translate the results of visual analysis into operable physical quantities.

[0081] S3.2: Calculate the energy weighting center of the weight distribution in physical coordinates as the equivalent optical coupling reference center.

[0082] Based on the above-established set of physical coordinate points and their corresponding weight distribution Calculate the energy weighting center under this weight distribution; this center is defined as the equivalent optical coupling reference center. The physical significance of this center lies in the fact that it reflects the energy centroid of the spatial projection of the real optical mode field onto the end face, rather than the centroid of its geometry.

[0083] coordinates The formula is obtained by using the weighted average method, as follows: ; ; because The result has been normalized within the valid region, and the theoretical denominator is 1. The denominator form is retained to accommodate minor deviations that may exist in actual engineering calculations. coordinates and The coordinate unit is micrometers.

[0084] To assess whether the reference center is sufficiently representative, i.e., whether light energy is concentrated in its vicinity, a local energy percentage index needs to be introduced. The method for calculating this index is as follows: with the equivalent optical coupling reference center... Centered on a circle, a pre-defined constant radius is used to assess energy concentration. (For example Within a circular region with radius , sum all weights. The sum of these values ​​is the local energy proportion index. . The larger the value, the more concentrated the light energy is near the reference center. In engineering, this is recommended... The value should be no less than 0.30 to ensure that the engineering orientation of the center point is clear and effective.

[0085] S3.3: Calculate the weighted variance of the weight distribution and synthesize the equivalent uncertainty radius.

[0086] In the complex end-face structure of thin-film lithium niobate chips, even if the weighted center is calculated, aiming at a single point may not be stable due to the potential broadening or multi-peak characteristics of the optical field distribution. Therefore, it is necessary to quantify the spatial uncertainty of the reference center and express it as an alignment target domain with a tolerance range, rather than an absolute point.

[0087] Weight distribution based on physical coordinates It can be calculated relative to the reference center. The weighted second moment, i.e., the weighted variance of the horizontal and vertical directions. and The calculation formula is as follows: ; ; Subsequently, the variances in these two directions are combined to form an equivalent uncertainty radius. This is used to describe the spatial dispersion of the weight distribution: ; The unit is micrometer. Its physical meaning is that the effective energy of the light field is mainly distributed within a region of approximately [micrometer value missing]. Center, with Within the region defined by the characteristic radius. In engineering practice, The value is typically between 0.2 micrometers and 3 micrometers; the more concentrated the light emission from the end face, the better. The smaller the size, the more dispersed the light field is due to the structure. The larger.

[0088] Equivalent optical coupling reference center coordinates Its uncertainty radius Packaging, forming an end-face space aligned with the target. This expression transforms the abstract optical center into an executable alignment target containing spatial tolerance information, effectively avoiding repeated adjustments or oscillations caused by pursuing an unstable theoretical point in subsequent alignment control.

[0089] S3.4: Map the equivalent optical coupling reference center and uncertainty radius to the assembly platform coordinate system.

[0090] The aforementioned calculations yielded the target alignment. Physical coordinate system located on the end face itself In the process, the assembly platform that ultimately controls the movement of the optical fiber has its own independent motion coordinate system. Therefore, the equivalent optical coupling reference center must be transformed to the platform coordinate system through coordinate mapping so that the motion control system can use it directly.

[0091] This mapping is achieved through a pre-performed system calibration. The calibration process determines the linear transformation relationship between the end face physical coordinates and the platform coordinates, which typically includes a scaling factor. and zero-point offset .

[0092] Using these calibration parameters, through a pre-calibrated linear relationship , The equivalent optical coupling reference center coordinates Transform to the assembly platform coordinate system to obtain the optical coupling reference center in the assembly platform coordinate system. .in, , The unit is usually micrometer. Scale factor The value range is generally between 0.8 and 1.2, used to absorb proportional differences caused by clamping tilt or incomplete alignment of coordinate axes; offset This depends on the platform's zero-point definition and the position of the imaging field of view.

[0093] Meanwhile, the uncertainty radius of the end face It is also necessary to consider the combined influence of the axial scale coefficients of the platform coordinate system to obtain the assembly tolerance radius in the assembly platform coordinate system. Considering that the two axes of the platform may have different scale factors, the synthesis formula is as follows: ; The unit is micrometer, which defines the permissible distance around the target point from the center of the fiber optic core in the platform coordinate system. The equivalent deviation radius.

[0094] This step yields the optical coupling reference center in the assembly platform coordinate system. and assembly tolerance radius This provides a unique, stable spatial reference quantity with clear physical tolerance for subsequent fiber optic pose alignment control. This reference quantity is rooted in the actual optical field and structural analysis, fundamentally avoiding the systematic deviations caused by traditional geometric center references.

[0095] In the technical solution of this disclosure, the effective region of the mode field in pixel space is transformed into a physical scale coordinate field. The energy weighting center of its light intensity distribution is calculated as the equivalent optical coupling reference center, and its equivalent uncertainty radius is synthesized by calculating the weighted variance. Finally, this center is mapped to the assembly platform coordinate system. This process transforms the abstract optical mode field energy centroid into a spatial alignment target with definite physical coordinates and tolerance radius in the platform coordinate system, thereby providing a reference quantity for direct motion control. This reference quantity originates from the comprehensive analysis of the real light field and structure, replacing the traditional geometric center reference and fundamentally eliminating the systematic alignment deviation caused by the decoupling of the two.

[0096] S4: Using the optical coupling reference center as the target, adjust the position and orientation of the optical fiber through feedback control so that the center of the fiber core approaches and aligns with the optical coupling reference center.

[0097] Figure 2 The fiber optic pose feedback control alignment flowchart provided in the embodiments of this disclosure is as follows: Figure 2 As shown, the alignment control process targets the optical coupling reference center generated in step S3. It adjusts the position and orientation of the optical fiber through feedback control, bringing the fiber core center close to and aligned with the reference center, achieving sub-micron level alignment. This is specifically achieved through the following sub-steps.

[0098] S4.1: Obtain the current position of the fiber spindle center and calculate its alignment deviation with the optical coupling reference center in the assembly platform coordinate system.

[0099] Alignment control begins with precise perception of the current spatial state of the optical fiber. First, a machine vision system performs real-time imaging and processing of the clamped fiber end face to identify and calculate the center coordinates of the fiber core in the assembly platform coordinate system at the current moment, denoted as . .

[0100] Next, the current position of this optical fiber is compared with the optical coupling reference center in the assembly platform coordinate system determined in step S3.4. By comparison, the two-dimensional deviation used to guide the platform's translational motion is calculated, namely the deviation in the X-axis direction. Deviation in the Y-axis direction ,in, and The unit is micrometers, which represent the amount of displacement that the optical fiber needs to compensate for on the X and Y axes of the platform, respectively.

[0101] To comprehensively assess the current overall offset, the current radial deviation is calculated. That is, the Euclidean distance between the current position of the optical fiber and the target reference center. . Also measured in micrometers, its value intuitively reflects the urgency of translational alignment and the required magnitude of movement, serving as a key basis for subsequently determining the translational control strategy. This step transforms the abstract alignment target into specific, executable motion error commands, ensuring that the entire alignment process closely revolves around the real optical coupling reference.

[0102] S4.2: Adaptively adjust the step size according to the magnitude of the deviation and perform piecewise translation approximation.

[0103] To avoid problems such as mechanical overshoot, platform oscillation, or slight loosening of the clamping mechanism due to excessive one-time displacement, a staged, variable-step control strategy is adopted for the translational approximation process. This strategy is based on the current radial deviation. The size of the scale is adaptively adjusted to change the step size, achieving a smooth transition from rapid coarse adjustment to fine-tuning.

[0104] In specific implementation, firstly according to The value determines the currently applicable step size coefficient. Its value selection rule is usually set as follows: when At that time, it was in the rough adjustment stage, taking The value is 0.5 to 0.8 to quickly approach the target; when At that time, we entered the fine-tuning stage and took... It approximates stably from 0.2 to 0.5; when At that time, it was in the fine-tuning stage, and the following steps were taken. The value is set to 0.05 to 0.2 for final submicron-level positioning.

[0105] Based on the calculated deviation and step size coefficient Generate the translation command that the platform should execute within the current control cycle. :in, , .

[0106] After the platform control system executes this command, the position of the fiber optic core center is updated to... ; Subsequently, the straight-line distance between the fiber core center and the optical coupling reference center was immediately recalculated based on the updated position, and this distance was taken as the residual radial deviation. .

[0107] This segmented approximation method, while ensuring alignment efficiency, minimizes the dynamic errors of the mechanical system, enabling the optical fiber to smoothly and accurately enter a circle centered on the optical coupling reference center and within the assembly tolerance radius. A circular region defined by a radius.

[0108] S4.3: Calculate and adjust fiber pitch and yaw attitude angles based on the symmetry of the light output distribution at the end face.

[0109] After aligning the fiber core to bring it close to the target, a slight pitch or yaw angle mismatch may still exist between the fiber and the chip waveguide. In the case of a waveguide endface with a high refractive index step, this mismatch can lead to a horizontal or vertical asymmetry in the spatial distribution of the emitted light field. Even with coordinate alignment, the coupling efficiency will remain suboptimal. Therefore, this sub-step quantitatively assesses the angle mismatch by analyzing the symmetry of the light distribution emitted from the chip endface and fine-tunes the fiber attitude accordingly.

[0110] The attitude adjustment is based on the physical coordinate weight distribution that has been fixed in steps S2 and S3. Reference center coordinates and physical coordinate point set First, calculate the x-coordinate of the reference center in the physical coordinate system. Weights for the right and left regions of the dividing line. The sum of the two, minus the sum of the two, gives the energy difference between the left and right sides. Similarly, using the vertical axis Using this as the dividing line, calculate the difference in the sum of the weights of the upper and lower regions to obtain the energy difference between them. .

[0111] To eliminate the influence of the total energy magnitude, the energy difference is normalized to obtain a dimensionless index characterizing the degree of asymmetry. and : ; ; Under ideal alignment, the light field distribution should be substantially symmetrical about the reference center. and It should be close to zero. If it is not zero, it indicates an angular mismatch. Based on the stated asymmetry... and The sign and magnitude of the value are used to calculate the required pitch angle change. and fiber optic yaw angle change : , ,in, and The unit is degrees. This is the angular scaling factor, typically ranging from 0.1 to 2.0. The specific value needs to be calibrated based on the angular resolution of the motion platform and the system sensitivity. The formula shows that the magnitude of the asymmetry determines the adjustment range, and its sign determines the direction of adjustment. A negative sign in the formula indicates that the adjustment direction is to reduce asymmetry, i.e., if... If the value is positive, the yaw angle needs to be adjusted in the negative direction. ,vice versa.

[0112] After performing attitude fine-tuning, since changes in fiber angle may slightly affect the projected position of its end face, it is necessary to re-acquire and calculate the translational residual deviation, denoted as . This step transforms the microscopic angular mismatch, which is difficult to observe directly, into a symmetry problem that can be analyzed through light field images, and enables feedback-based correction, which is crucial for improving the final coupling efficiency.

[0113] S4.4: Based on the translation tolerance and the light field asymmetry threshold, determine that the alignment is complete and output the final pose.

[0114] The alignment process requires clear and quantifiable criteria to conclude, rather than subjective judgment. The criteria must simultaneously meet the requirements of translational accuracy and attitude accuracy, ensuring that the optical fiber is in a stable pose with good angle matching and targeting the optical coupling reference center.

[0115] First, translation residual deviation It must be less than or equal to the assembly tolerance radius determined in step S3.4. This means that the center of the fiber optic core has entered the permissible alignment target area.

[0116] Secondly, the absolute value of the optical field asymmetry evaluated after attitude adjustment. and It must be less than the preset allowable asymmetry threshold. . Typically, the value is between 0.02 and 0.10, representing the level of optical field asymmetry corresponding to the maximum angular deviation that the system can accept without affecting the coupling performance.

[0117] If and only if the above three conditions (i.e.) , and When both conditions are met, alignment is considered complete, and a completion determination flag is set. Set to 1; if any condition is not met, then... Set to 0.

[0118] After alignment is complete, the system outputs the final, stable fiber pose state. This status includes the final coordinates of the fiber optic core center platform. And the cumulative attitude angle of the optical fiber after adjustment. This output pose state is the direct basis for subsequent locking and encapsulation operations, indicating that the active control of the fiber pose has successfully achieved the engineering goals set by the technical solution.

[0119] In the technical solution of this disclosure, the optical coupling reference center and its tolerance radius are used as a reference. The fiber pose is acquired in real time and the alignment deviation is calculated. Segmented translational approximation is performed to stably update the fiber position. Then, based on the symmetry of the light output distribution at the end face, attitude fine-tuning is performed to suppress angular mismatch. Finally, alignment is determined to be complete based on the translation tolerance and asymmetry threshold. This feedback control method achieves a smooth transition of fiber pose from macroscopic coarse adjustment to submicron-level fine adjustment, and simultaneously optimizes both translation and angular degrees of freedom, enabling the fiber core to be stably aligned to the optimal coupling position in an optical sense, thereby significantly improving alignment accuracy and final coupling efficiency.

[0120] S5: After alignment is completed, evaluate and lock the stable fiber pose, and output the final coupling pose and performance indicators.

[0121] After alignment is completed in step S4, the stability of the alignment state is evaluated and confirmed, the optimal pose parameters are locked, and performance indicators such as insertion loss are measured to complete the final feedback adjustment and locking of the alignment process. This is achieved through the following sub-steps.

[0122] S5.1: Evaluate pose fluctuations within a continuous sampling period and determine a stable coincident state.

[0123] Based on the determination of single alignment completion in step S4, to ensure that this alignment is not instantaneous or accidental, but a stable state that can be maintained continuously, it is necessary to continuously monitor and quantitatively evaluate the alignment stability over a period of time. A single sampling entering the tolerance range is insufficient to prove stability; the pose fluctuations must remain within the strictly permissible range for engineering purposes over multiple consecutive sampling periods.

[0124] In practice, a suitable sampling period should be set. Its value typically ranges from 10 milliseconds to 200 milliseconds to accommodate the platform's dynamic response characteristics and image processing frame rate. Simultaneously, the number of consecutive samples N is set, generally between 5 and 50, to ensure the statistical reliability of the evaluation results.

[0125] At each sampling time i, the system acquires the coordinates of the fiber core center in the platform coordinate system in real time. And calculate the distance between the fiber core center and the optical coupling reference center at that moment. The straight-line distance between them is used as the instantaneous radial deviation. After collecting N data points consecutively, calculate N instantaneous radial deviations. The root mean square value is used as a stability evaluation metric. , The unit is micrometers. The smaller the value, the smaller the fluctuation amplitude of the fiber center around the reference center during the observation period, and the better the repeatability and stability of the pose.

[0126] To determine whether a stable coincidence state has been achieved in an engineering sense, it is necessary to... With a preset stability threshold Compare them. The value range is typically set to 0.1 micrometers to 0.8 micrometers, and its value must be less than the assembly tolerance radius. To ensure a stable state, the requirements are more stringent than those for basic focus completion. The stable alignment state must meet the following two conditions: (a) the single alignment completion flag is true, i.e. (b) The stability evaluation metrics meet the requirements, i.e. The system determines that it has reached a stable coincidence if and only if conditions (a) and (b) are both true, and sets the stability determination flag accordingly. Set to 1 otherwise set to 0. This step transforms the abstract stability requirement into measurable and comparable engineering parameters by quantitatively analyzing the positional fluctuations in the time series, effectively avoiding false locking caused by instantaneous disturbances, system noise, or mechanical drift.

[0127] S5.2: Lock the stable relative pose of the optical fiber, smooth and average it, and then solidify it into the final set of coupling parameters.

[0128] Once the stability determination flag is established... If the value is confirmed as 1, the system will immediately perform a locking operation. The core of locking is reading and saving the pose parameters of the fiber relative to the chip, which is currently in an optimal and stable state, forming a parameter set that can be directly called and inherited by subsequent packaging process steps. To prevent the locked value from being affected by random noise or instantaneous jitter from a single sampling, the locking parameters are obtained by smoothing and averaging N consecutive sampling data to filter out fluctuations and obtain a more representative true pose value. In this embodiment, an arithmetic average is used to achieve the smoothing average.

[0129] First, calculate the locking value of the fiber core center coordinates. The calculation formula is as follows: ; ; Simultaneously, the locked value of the fiber attitude angle is calculated. The calculation formula is as follows: ; ; in, and The cumulative pitch and yaw angles of the optical fiber recorded at the i-th sampling time are denoted as .

[0130] Once the locking operation is successfully executed, the system generates a lock execution flag. And set it to 1. If locking is not performed due to reasons such as inadequate stability or external intervention, then If the value remains at 0, the alignment control process will remain in the adjustment state or trigger an alarm.

[0131] The final set of coupling alignment parameters and lock mark This parameter set is stored in the system's non-volatile memory or transmitted to the upper control system in real time. It contains not only spatial position information with sub-micron precision but also angular information with sub-degree precision, providing a precise and traceable pose reference for subsequent permanent encapsulation processes such as UV curing adhesive dispensing, laser welding, or mechanical pressing. This ensures that the aligned optimal optical coupling state can be completely maintained and reproduced in subsequent processing.

[0132] S5.3: Measure and calculate the average insertion loss, and output the final result including pose parameters and performance indicators.

[0133] During or after pose locking, in order to objectively verify the effectiveness and superiority of this alignment from the perspective of optical transmission performance, it is necessary to measure and calculate the final insertion loss index. Insertion loss directly and quantitatively reflects the degree of power attenuation of the optical signal during the coupling process from the chip waveguide into the optical fiber, and is the ultimate physical quantity for realizing and verifying low-loss coupling.

[0134] First, while maintaining the locked pose absolutely unchanged, a stable optical signal of known power is injected into the input terminal of the thin-film lithium niobate chip. The output optical power is simultaneously measured at the output end of the optical fiber using a calibrated optical power meter. Input and output power must use the same unit, typically milliwatts or decibel-milliwatts.

[0135] To obtain stable measurement values ​​unaffected by instantaneous fluctuations, the output optical power readings are also synchronously read over N consecutive sampling periods. The average value of N readings is taken as the output power. Based on average input power and average output power Calculate the insertion loss of this alignment and coupling. The unit is decibels: ; In engineering practice, an acceptance threshold can be set based on the specific device type, operating wavelength, and coupling method. Its value typically ranges from 0.5 dB to 5 dB. The system will calculate the... It is compared with this threshold, but regardless of the comparison result, i.e., regardless of Regardless of whether this threshold is met, the system will output the final result of this alignment completely and objectively. .

[0136] Final result It is a composite structure that integrates all key process and performance data, including: a locked set of alignment parameters. Insertion loss obtained by measurement and calculation and lock execution flag .Right now .

[0137] This result not only includes the mechanical pose parameters required for high-precision spatial alignment but also provides optical performance indicators that directly reflect coupling efficiency, forming a complete technical loop from machine vision positioning, intelligent analysis and decision-making, motion control execution to optical performance verification. This final result will be recorded by the system and can be used to generate detailed test reports, achieve process traceability, or directly guide subsequent packaging processes. This signifies that the thin-film lithium niobate chip fiber-optic coupling machine vision alignment control method described in this invention has been successfully implemented and has achieved its expected engineering goals, providing crucial technical support for the manufacturing of high-performance, high-reliability optical devices.

[0138] In the technical solution of this disclosure embodiment, after alignment is completed, a stable coincidence criterion, more stringent than that for a single alignment completion, is established by continuously monitoring instantaneous pose fluctuations and calculating stability evaluation quantities, thereby ensuring the time-dependent continuity and reliability of the obtained excellent coupling state. Subsequently, the system locks and smoothly averages the final coupling alignment parameter set, while simultaneously measuring and calculating the average insertion loss index, forming a complete final result including pose parameters and performance indicators. This step realizes a complete control process from dynamic alignment to static locking, and the output parameter set provides an accurate and traceable pose reference for subsequent permanent packaging, ensuring that the optimized optical coupling state can be maintained and reproduced throughout the entire device manufacturing process.

[0139] According to embodiments of this disclosure, an electronic device is also provided, which may include a processor, a communications interface, a memory, and a communication bus, wherein the processor, the communications interface, and the memory communicate with each other via the communication bus. The processor can invoke logical instructions stored in the memory to execute the methods provided in the above embodiments.

[0140] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0141] On the other hand, this disclosure also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the methods provided in the above embodiments.

[0142] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0143] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0144] It should be understood that the above embodiments are only used to illustrate the technical solutions of this disclosure, and not to limit them; although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure.

Claims

1. A method for aligning and controlling machine vision using a thin-film lithium niobate chip coupled with fiber optics, characterized in that, The method includes: S1. Acquire high-resolution images of the waveguide end face of the thin-film lithium niobate chip, extract the etched step contour, sidewall boundary, and normalized light intensity distribution from them to form a set of multi-structure visual representations. S2. Based on the multi-structure visual representation set, and combining the emitted light intensity distribution with the geometric constraints provided by the sidewall boundaries, determine the effective area of ​​the mode field. S3. Based on the light intensity distribution within the effective area of ​​the mode field, calculate its energy weighting center and map the energy weighting center to the assembly platform coordinate system to obtain the optical coupling reference center for alignment. S4. Using the optical coupling reference center as the target, adjust the position and attitude of the optical fiber through feedback control so that the center of the fiber core approaches and aligns with the optical coupling reference center. S5. After alignment is completed, evaluate and lock the stable fiber pose, and output the final coupling pose and performance indicators. S3 includes: The effective area of ​​the mode field and its corresponding normalized light intensity weight distribution are transformed from image pixel coordinates to end face physical coordinates; In the physical coordinates of the end face, its energy weighting center is calculated based on the transformed weight distribution and used as the equivalent optical coupling reference center; In the physical coordinates of the end face, the equivalent uncertainty radius characterizing the spatial discreteness of the optical field is calculated based on the transformed weight distribution; Based on the equivalent optical coupling reference center and the equivalent uncertainty radius, the end face spatial alignment with the target is formed; The end face space alignment target is mapped to the assembly platform coordinate system to obtain the optical coupling reference center and assembly tolerance radius in the platform coordinate system used for direct alignment control.

2. The fiber-coupled machine vision alignment control method for thin-film lithium niobate chips according to claim 1, characterized in that, S1 includes: The grayscale image of the waveguide end face is acquired by a microscopic imaging system, the pixel coordinate system of the image is used as a unified spatial reference frame, and the conversion factor from pixel to physical length is calculated. In the pixel coordinate system, the etched step contour is extracted based on gradient operation and threshold segmentation, and the sidewall boundary of the waveguide is extracted based on the step contour through local constraint search. Within the area defined by the sidewall boundary, the pixel values ​​of the grayscale image are normalized to obtain the normalized light intensity distribution, and its energy weighting center is calculated. The etched step contour, sidewall boundary, normalized light intensity distribution, and energy weighting center are integrated in the pixel coordinate system to form a multi-structure visual representation set.

3. The fiber-coupled machine vision alignment control method for thin-film lithium niobate chips according to claim 1, characterized in that, S2 includes: Based on the waveguide sidewall boundary in the multi-structure visual representation set, an effective analysis window for the end face is delineated in the image. Within the window, based on the normalized light intensity value of each pixel, the energy coverage ratio threshold method is used to filter out the candidate set of light-emitting core areas whose cumulative energy reaches a preset ratio; Based on the area covered by the candidate set of the light-emitting core region, the minimum distance from each pixel in the window to the sidewall boundary and the etched step contour is calculated, and a structural weight function is constructed based on the distance. The normalized emitted light intensity distribution within the window is corrected using the structure weight function to obtain the structure-related corrected emitted light distribution. The final effective mode field region and its boundary are then determined using the energy coverage ratio threshold method.

4. The fiber-coupled machine vision alignment control method for thin-film lithium niobate chips according to claim 1, characterized in that, The calculation of the equivalent uncertainty radius characterizing the spatial discreteness of the light field based on the transformed weight distribution includes: Calculate the lateral weighted variance and longitudinal weighted variance of the weight distribution in the physical coordinates of the end face relative to the equivalent optical coupling reference center; The equivalent uncertainty radius is synthesized from the horizontal weighted variance and the vertical weighted variance.

5. The fiber-coupled machine vision alignment control method for thin-film lithium niobate chips according to claim 1, characterized in that, S4 includes: Real-time acquisition of the current position coordinates of the fiber optic core center in the assembly platform coordinate system; Calculate the positional deviation between the current position and the optical coupling reference center in the platform coordinate system; Based on the magnitude of the positional deviation, a segmented and step-adaptive strategy is adopted to control the actuator to drive the optical fiber to translate and approach, so that the center of the optical fiber core enters the area defined by the optical coupling reference center as the center and the assembly tolerance radius as the radius.

6. The fiber-coupled machine vision alignment control method for thin-film lithium niobate chips according to claim 5, characterized in that, After the process of bringing the fiber core center into the region defined by the optical coupling reference center and the assembly tolerance as the radius, it further includes: Calculate the total light intensity in the left and right regions bounded by the horizontal coordinate of the optical coupling reference center, and the upper and lower regions bounded by the vertical coordinate, and calculate the left-right energy difference and the upper-lower energy difference. Divide the energy difference by the total emitted light intensity within the entire effective area of ​​the mode field to obtain the normalized left-right asymmetry and up-down asymmetry. The yaw angle adjustment is calculated based on the left-right asymmetry, the pitch angle adjustment is calculated based on the up-down asymmetry, and attitude fine-tuning is performed accordingly. When the adjusted translational residual deviation is less than the assembly tolerance radius, and the absolute values ​​of the adjusted left-right and up-down optical field distribution asymmetry are both lower than the preset threshold, the fiber pose alignment is determined to be complete.

7. The fiber-coupled machine vision alignment control method for thin-film lithium niobate chips according to claim 1, characterized in that, S5 includes: After the alignment is determined, the instantaneous distance fluctuation between the fiber core center and the optical coupling reference center is monitored over multiple consecutive sampling periods. Calculate the root mean square value of the consecutive instantaneous distances as a stability evaluation metric; When the stability evaluation value is less than or equal to a preset stability threshold, it is determined that a stable coincidence state has been reached. For fiber pose when a stable coincidence state is reached, a locking operation is performed. The fiber core center coordinates and attitude angles continuously sampled during the stable state are smoothed and averaged, and then solidified into the final coupling alignment parameter set. Under the pose defined by the final coupling alignment parameter set, the input power of the optical signal and the output power after optical fiber coupling are measured, and the average insertion loss is calculated. The output contains the final result including the final set of coupling alignment parameters and the average insertion loss.

8. An electronic device, characterized in that, The electronic device includes a memory and at least one processor, the memory storing a computer program, and the processor executing the computer program to implement the thin-film lithium niobate chip fiber-coupled machine vision alignment control method according to any one of claims 1-7.

9. A computer storage medium, characterized in that, It stores a computer program, which, when executed, implements the thin-film lithium niobate chip fiber-coupled machine vision alignment control method according to any one of claims 1-7.

Images