Zoom lens module active calibration and multi-degree-of-freedom coupling debugging device and system

Abstract

The present application relates to optical imaging module assembly and active calibration control technical field, specifically relates to zoom lens module active calibration and multi-degree of freedom coupling debugging device and system, the device contains positioning clamping, double action platform, imaging evaluation, displacement measurement, cooperative control; fixed second component is the reference, platform three direction translation three direction rotation to first / third component; evaluation output image quality state index; solidification stage measurement online estimates coupling pose error; control active calibration output calibration instruction, solidification output trace compensation and irradiation energy regulation, error convergence or solidification reaches criterion stop compensation, with second component as reference, six degrees of freedom pose of double action platform; with image quality state index closed loop active calibration; solidification period online measurement coupling pose error, joint trace compensation and adjust UV energy, reach convergence / solidification criterion stop compensation, improve final state image quality consistency, yield and process robustness.

Description

Technical Field

[0001] This invention relates to the field of optical imaging module assembly and active calibration control technology, and more specifically, to a device and system for active calibration and multi-degree-of-freedom coupling debugging of zoom lens modules. Background Technology

[0002] In the assembly and manufacturing process of zoom lens modules, it is usually necessary to establish a stable assembly benchmark for the lens components, base components, and image sensor components. Pose adjustments are then used to achieve optical axis alignment, controllable tilt angles, and image plane position matching, thereby obtaining imaging results that meet resolution, contrast, and off-axis consistency requirements. With the miniaturization and high-pixel count of products, the assembly tolerance window is constantly narrowing. Traditional methods relying on mechanical positioning and experience-based adjustments are insufficient to balance efficiency and consistency. Therefore, the industry is gradually adopting an active calibration process based on imaging results: during the adjustment process, images are acquired in real time and evaluation values ​​are calculated. The pose is then iteratively adjusted until the target imaging quality requirements are met. Subsequently, ultraviolet light is used to cure adhesive materials or solidify the structure to achieve locking, forming the final assembly state.

[0003] In existing active calibration and curing locking processes, common practices include: First, using a multi-degree-of-freedom platform to perform three-way translation and three-way rotation adjustment on lens components or image sensor components to obtain an assembly pose closer to the optimal imaging state; Second, performing closed-loop adjustments based on evaluation metrics such as image sharpness, contrast, or frequency domain response to improve imaging quality and assembly yield; Third, using fixed irradiance or fixed irradiation time during the curing stage to meet the requirements for adhesive layer curing strength and cycle time; Fourth, some processes are supplemented with metrological methods such as displacement, angle, or interference phase to monitor pose changes during the curing process. While the aforementioned technical approach is usable in engineering practice, it may still encounter some issues requiring further coordination in zoom module scenarios with high consistency requirements. For example, if image evaluation focuses on the field center while neglecting off-axis regions, it may lead to uneven image quality at edges or corners after curing. Furthermore, adhesive shrinkage, thermal effects, or stress release during UV curing may cause slow drift, causing the actual pose at the end of curing to deviate from the optimal pose before curing. Additionally, the changes in the lens components relative to the base are coupled with the changes in the image sensor components relative to the base; monitoring only a single relative link or relying solely on a single information source may make it difficult to reliably separate translation and rotation errors and achieve stable compensation. Moreover, irradiation energy parameters and pose stability are mutually influential; without an adjustment strategy linked to image quality and pose state, it may be difficult to simultaneously achieve sufficient curing and stable image quality. Therefore, establishing a continuous, observable, and controllable closed-loop collaborative process between the active calibration stage and the UV curing stage, ensuring that image quality remains within a preset range after curing, has become an important direction for optimizing related manufacturing equipment and processes. Summary of the Invention

[0004] In view of the shortcomings of the existing technology, the purpose of this invention is to provide an active calibration and multi-degree-of-freedom coupling debugging device and system for zoom lens modules.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The zoom lens module active calibration and multi-degree-of-freedom coupling debugging device includes: Positioning and clamping module, dual-motion platform module, imaging evaluation module, displacement measurement module, and collaborative control module; The positioning and clamping module is used to fix the second module component and provide an assembly reference. The dual-motion platform module is used to support the first module component and the third module component respectively, and to achieve multi-degree-of-freedom pose adjustment of three-way translation and three-way rotation relative to the second module component; The imaging evaluation module is used to acquire imaging data output by the third module component during the active calibration phase and the ultraviolet curing phase, and output dynamic indicators of image quality. The displacement measurement module is used to estimate the relative pose between the first module component and the second module component, and between the third module component and the second module component, during the UV curing stage, and output the coupled relative pose error state quantity. The collaborative control module connects to the dual-motion platform module and the ultraviolet curing light source. The collaborative control module is used for: During the active calibration phase, pose calibration commands for the dual-motion platform module are output based on dynamic image quality indicators. During the UV curing stage, the dual-motion platform module outputs a micro-pose compensation command based on the dynamic index of image quality and the coupled relative pose error state quantity. During the ultraviolet curing stage, an irradiation energy adjustment command is output to the ultraviolet curing light source; When the coupled relative pose error state quantity meets the preset convergence criterion or the preset solidification criterion, the output of the micro pose compensation command stops.

[0006] Furthermore, the first module component is a zoom lens barrel assembly, the second module component is a module base assembly, and the third module component is an image sensor assembly.

[0007] Furthermore, the positioning and clamping module includes a reference platform, at least three non-collinear positioning reference elements, and a clamping mechanism. The positioning reference elements include at least two of the following: a reference plane, a positioning pin, and a stop surface. The second module component has a reference surface and / or a reference hole that mates with the positioning reference elements. After mating, the clamping mechanism presses or adsorbs the second module component onto the reference platform to establish an assembly reference.

[0008] Furthermore, the imaging evaluation module extracts and normalizes multi-region image quality features based on the imaging data within a preset time window; the normalized multi-region image quality features constitute a feature vector, which is then fused by a weight matrix and combined with temporal filtering and trend penalty processing to form a dynamic index of image quality.

[0009] Furthermore, the multiple regions include a central region of the field of view, at least one inner ring region, at least one outer ring region, and an edge region. The edge region includes at least one type of four-corner sub-region and edge midpoint sub-region. The weights of each region are configured according to the field of view position and image quality sensitivity, and are used for weight matrix mapping and fusion.

[0010] Furthermore, image quality features include one or more of the following: modulation transfer function, spatial frequency response, contrast, astigmatism measure, and field curvature measure.

[0011] Furthermore, the online estimation is based on the synchronization and calibration of the observation time, and establishes a state space model by combining phase unwrapping, rigid body motion constraints and multi-source fusion nonlinear state estimation; the output is the coupled relative pose error state quantity, which includes the joint state of the six-degree-of-freedom relative pose deviation and its dynamic terms.

[0012] Furthermore, the collaborative control module is used for: During the active calibration phase, the image quality dynamic index is determined as the target value, and a pose calibration command is generated based on the deviation between the image quality dynamic index and the target value to drive the dual-motion platform module to adjust the pose of the first module component and / or the third module component so that the image quality dynamic index approaches the target value. During the UV curing stage, the model prediction extrapolates the predicted coupled pose error state quantity based on the deviation of the image quality dynamic index relative to the target value and the coupled relative pose error state quantity. Then, pose prediction compensation trajectories for the first module component and the third module component are generated, and micro pose compensation commands are generated and output based on the pose prediction compensation trajectories.

[0013] Furthermore, the irradiation energy adjustment command is used to adjust at least one of the irradiance, irradiation duration, and duty cycle, so that the dynamic indicators of image quality are kept within a preset range when curing is completed.

[0014] Compared with the prior art, the present invention has the following beneficial effects: This invention provides a stable assembly reference for the second module component through positioning and clamping. A dual-motion platform carries the first and third module components respectively, enabling three-way translation and three-way rotation relative to the second module component. This covers key sources of assembly freedom such as optical axis eccentricity, tilt error, and image plane position error. Simultaneously, during the active calibration phase, a dynamic image quality index based on imaging data is introduced, and pose calibration commands are output accordingly. This transforms pose optimization from traditional experience-based adjustment to an iterative process driven by a closed-loop index, achieving repeatable and quantifiable calibration convergence under the same assembly reference. This improves batch consistency and assembly efficiency, and reduces discrete fluctuations introduced by different operating habits. The dynamic image quality index comprehensively represents both the center and off-axis regions of the field of view, avoiding edge image quality unevenness caused by relying solely on single-point sharpness, making it easier to achieve image uniformity of the zoom lens module at different focal lengths.

[0015] In the UV curing stage, displacement measurement is introduced to estimate the relative poses between the first and second module components, and between the third and second module components, online. The system outputs a coupled relative pose error state quantity containing six-degree-of-freedom relative pose deviations and their dynamic terms. This allows for the continuous characterization of slow drifts and short-term disturbances caused by adhesive shrinkage, thermal effects, or stress release during curing, and enables the differentiation of the coupling effects between two relative links. Based on this, the image quality dynamic index and the coupled relative pose error state quantity are combined to generate micro-pose compensation commands. This forms a collaborative control path that corrects image quality deviations and counteracts pose drift trends, preventing the optimal pose reached before curing from being "pulled off" at the final curing state, thereby significantly improving imaging stability and yield upon completion of curing.

[0016] During the UV curing stage, an irradiation energy adjustment command is output to the UV curing light source to adjust at least one of the following: irradiance, irradiation duration, and duty cycle. This binds the curing process to the image quality maintenance target. On the one hand, the curing rate and shrinkage rate are adjusted by controlling the energy input, reducing the growth slope of pose drift. On the other hand, with micro-pose compensation, the dynamic indicators of image quality are kept within a preset range when curing is completed, expanding the process window and improving process robustness. Simultaneously, this invention sets a preset convergence criterion or a preset curing criterion. When the coupled relative pose error state quantity reaches the convergence requirement or the curing conditions are met, the output of the micro-pose compensation command is stopped. This effectively avoids the risks of vibration, overcompensation, and springback introduced by continuous movement at the curing end, making the final locked state more stable and repeatable, and facilitating consistent final state acceptance under production line cycle time. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall structure and connection relationship of the zoom lens module active calibration and multi-degree-of-freedom coupling debugging device of the present invention. Figure 2 A schematic diagram of the structure for fixing the second module component and establishing an assembly reference in the positioning and clamping module of the present invention; Figure 3 This is a schematic diagram illustrating the pose adjustment of the dual-motion platform module of the present invention, which carries the first module component and the third module component and achieves three-way translation and three-way rotation relative to the second module component. Detailed Implementation

[0018] Reference Figures 1 to 3 The zoom lens module active calibration and multi-degree-of-freedom coupling debugging device includes: This device is designed for two types of working conditions in the assembly process of zoom lens modules: "active calibration" and "ultraviolet curing". Through the collaboration of multiple modules, the first module component, the second module component, and the third module component maintain a controllable relative positional relationship during the assembly and curing process. It can also make timely adjustments and compensations when the imaging quality changes, thereby achieving multi-degree-of-freedom coupling debugging and stable curing.

[0019] Positioning and clamping module, dual-motion platform module, imaging evaluation module, displacement measurement module, and collaborative control module; The above modules constitute a complete closed-loop debugging system: the positioning and clamping module is responsible for establishing a stable assembly benchmark; the dual-motion platform module provides multi-degree-of-freedom pose adjustment capability; the imaging evaluation module converts "imaging data" into image quality dynamic indicators that can be used for control; the displacement measurement module provides coupled relative pose error state quantities as key state inputs during the solidification stage; and the collaborative control module associates the image quality dynamic indicators with the coupled relative pose error state quantities, outputting pose-type commands and energy adjustment-type commands to complete the two-stage collaboration.

[0020] The positioning and clamping module is used to fix the second module component and provide an assembly reference; the second module component serves as a "reference part" for assembly and debugging. By fixing the second module component, the device obtains a stable spatial reference, thereby ensuring that the pose adjustment of the first and third module components has consistent reference coordinates and assembly references, avoiding instability or reduced repeatability of calibration results due to reference drift.

[0021] In one specific embodiment, the second module component is placed on a reference platform and the assembly reference is established. Specifically, the reference surfaces and reference holes of the reference platform and the second module component are first cleaned. Then, the reference plane in the positioning reference element is made to fit with the reference surface of the second module component. The positioning pin in the positioning reference element is inserted into the reference hole of the second module component to complete the binary constraint of the plane and the hole. Then, the stop surface in the positioning reference element is made to contact the lateral reference surface of the second module component to form a third-direction limiting constraint, thereby achieving geometric positioning in at least three non-collinear directions. After the mating is completed, the clamping mechanism is driven to apply a clamping force or suction force to the second module component, so that the second module component obtains a stable and repeatable positioning state on the reference platform. The clamping force loading path keeps the second module component in contact with the reference plane and maintains the coaxial constraint of the positioning pin in the reference hole, avoiding loosening and displacement introduced by subsequent actions. For example, when the reference hole of the second module component has minor manufacturing tolerances, a locating pin with a lead-in fillet can be used in conjunction with the preload direction of the stop surface to allow the second module component to automatically fall into the defined constraint boundary after clamping. When the surface of the second module component is easily affected by clamping deformation, an adsorption fixing method can be adopted and a partitioned adsorption channel can be set on the reference platform to make the adsorption force evenly act on the stress area of ​​the second module component. This establishes an assembly reference without changing the geometric reference of the second module component, providing a stable reference for the subsequent positional adjustment of the first and third module components.

[0022] The dual-motion platform module is used to support the first and third module components respectively, and to achieve multi-degree-of-freedom pose adjustment in three-way translation and three-way rotation relative to the second module component. It provides multi-degree-of-freedom pose adjustment capabilities for both the first and third module components, and this adjustment is performed relative to the second module component. The dual-motion platform module supports the first and third module components respectively, enabling them to be decoupled or coupled within the three-way translation and three-way rotation range to meet the fine calibration requirements of multi-degree-of-freedom pose during assembly; simultaneously, it provides the actuator basis for micro-pose compensation in the subsequent curing stage.

[0023] In one specific embodiment, after fixing the module base assembly as an assembly reference to the tooling reference surface, the zoom lens barrel assembly and the image sensor assembly are respectively mounted on the first and third bearing seats of the dual-motion platform. Then, with the reference surface of the module base assembly as the zero-position reference, a multi-degree-of-freedom pose adjustment process of three-way translation and three-way rotation is performed: first, the optical axis of the zoom lens barrel assembly is roughly placed within the reference hole axis range of the module base assembly through three-way translation; then, the end face of the zoom lens barrel assembly is made to be parallel to the reference surface of the module base assembly through three-way rotation to reduce the initial eccentricity and tilt of the assembly. The same three-way translation and three-way rotation are performed on the image sensor assembly to make the photosensitive surface of the image sensor assembly and the image output surface of the zoom lens barrel assembly form an adjustable relative relationship in space. For example, when the zoom lens assembly exhibits inconsistent edge sharpness when imaging a target at infinity at the telephoto end, a small rotation around the horizontal and vertical axes can be applied to the image sensor assembly to correct the tilt angle, followed by a translation along the optical axis to compensate for the image plane position. When the center is sharp but the overall image is offset, a lateral translation can be applied to the zoom lens assembly to correct the eccentricity, followed by a small translation of the image sensor assembly to complete the alignment. Through this sequence of actions—coarse-to-fine, translation-to-rotation, and iteratively alternating—the poses of the two components relative to the module base assembly can be adjusted independently or coupled in all six degrees of freedom, thus creating a controllable initial assembly state for stable maintenance during subsequent active calibration and curing stages.

[0024] The imaging evaluation module acquires imaging data output from the third module during the active calibration and UV curing phases, and outputs dynamic image quality indicators. It transforms the imaging data output from the third module into comparable, traceable, and controllable indicators. During the active calibration phase, the dynamic image quality indicators are used to determine if the current pose meets the standards and guide the generation of pose calibration commands. Similarly, the UV curing phase requires continuous output of dynamic image quality indicators to characterize the dynamic changes in imaging quality during curing. This allows the control system to compensate for pose deviations and adjust irradiation energy based on the indicator deviations, ensuring that the imaging quality remains within a preset range upon completion of curing.

[0025] In one specific implementation, the third module outputs continuous imaging data, which is then segmented and processed by the imaging evaluation module within a preset time window. The preset time window can be set as a sliding window covering several frames of images to balance real-time performance and noise resistance. Within each time window, the imaging evaluation module first performs brightness and black level consistency processing on each frame of image, and then generates a multi-region mask according to a fixed field of view geometry division rule. The multi-region mask includes at least a central region of the field of view, at least one inner ring region, at least one outer ring region, and an edge region. The edge region further includes at least one of the four corner sub-regions and the edge midpoint sub-region, thereby ensuring that the sampling positions of each region within the same time window are consistent and repeatable, meeting the requirements of the active calibration stage and the ultraviolet curing stage for homogeneous comparison. For example, when the target is a checkerboard or oblique target, the central region is used to reflect the on-axis imaging state, the outer ring region is used to reflect the off-axis image quality, and the four corner sub-regions are used to amplify the changing trends of off-axis aberrations such as astigmatism and field curvature. After region segmentation, the imaging evaluation module extracts and normalizes multi-region image quality features for each region within a time window. These features include at least one or more of the following: modulation transfer function, spatial frequency response, contrast, astigmatism, and field curvature. Normalization can combine amplitude scaling based on historical steady-state intervals with variance constraints based on noise levels to unify different feature dimensions and make them insensitive to exposure fluctuations. The normalized multi-region image quality features are then concatenated by region and feature dimension to form a feature vector. A weight matrix is ​​configured to map and fuse the feature vector. The weights for each region are configured according to the field of view position and image quality sensitivity. For example, higher weights are given to the corner sub-regions to improve sensitivity to off-axis image quality drift, while stable weights are given to the central region of the field of view to avoid excessive tracking of local noise. Furthermore, when the zoom lens assembly is at the telephoto end, the weight of the outer ring region can be increased to highlight edge resolution consistency; at the wide-angle end, the weight of the midpoint sub-region at the edge can be increased to highlight the impact of distortion-related detail changes on image quality. After obtaining the mapped and fused results, the imaging evaluation module applies temporal filtering and trend penalty processing to the time window sequence to form a dynamic image quality index. Temporal filtering can use a recursive form to suppress short-term noise and maintain the phase response of the index. Trend penalty is used to impose an additional cost on persistent shifts to expose the gradual drift caused by the curing process earlier, so that the dynamic image quality index can reflect both the instantaneous image quality state and the slow evolution of the curing stage. In the active calibration stage, the dynamic image quality index is used to characterize the degree of immediate improvement brought about by the calibration action, which is convenient for iteratively searching for the optimal pose. In the ultraviolet curing stage, the dynamic image quality index is used to characterize whether the image quality remains stable during the curing process and remains detectable to small changes. For example, when the thermal effect caused by irradiation or the shrinkage of the adhesive layer causes the modulation transfer function of the edge region to slowly decrease, the trend penalty makes the index more sensitive to this decrease, so that risks can be identified earlier and subsequent compensation strategies can be used to ensure that the index remains within the preset range when curing is completed.

[0026] In one embodiment, the image quality dynamic index is defined and calculated as follows: Region Division: The field of view is divided using a normalized radius, with the geometric center of the imaging plane as a reference. The central region can be defined as a region with a normalized radius not greater than a preset center threshold; the inner ring region is defined as a region with a normalized radius between the center threshold and the inner ring threshold; the outer ring region is defined as a region with a normalized radius between the inner ring threshold and the outer ring threshold; and the edge region is defined as a region with a normalized radius greater than the outer ring threshold. The edge region can be further divided into four-corner sub-regions and edge midpoint sub-regions based on corner positions. The aforementioned thresholds can be preset according to the imaging size and the distribution of image quality sensitive areas. Preferably, the central region covers the effective area reflecting on-axis image quality, while the outer ring and edge regions cover the areas used to amplify off-axis aberration changes.

[0027] Feature extraction: Within each time window, image quality features are extracted for each frame according to the aforementioned regions. These image quality features may include at least one or more of the following: modulation transfer function correlation features, spatial frequency response correlation features, contrast features, astigmatism characteristics, and field curvature characteristics. When the target is a slanted target, the slanted method can be used to obtain the spatial frequency response; when the target is a checkerboard target, features can be constructed based on the frequency domain response or contrast response of the corner neighborhood.

[0028] Normalization: Dimensionless normalization is performed on each feature in each region to make different features comparable. Normalization can be achieved based on statistics of a steady-state reference interval: using the mean value of features in the pre-calibrated or actively calibrated stable phase as the reference center, and using the statistics of feature fluctuations in that phase as the scale, normalized features are obtained; at the same time, a suppression term for exposure fluctuations can be superimposed to reduce the impact of brightness changes on the index.

[0029] Fusion: The normalized multi-region image quality features are concatenated according to region and feature dimensions to form a feature vector, and then fused through weight mapping to obtain the window quality. The weights can be configured according to the field of view position and image quality sensitivity. It is preferable to assign higher weights to the four corner sub-regions to improve the sensitivity to off-axis image quality drift; and to assign stable weights to the central region to reduce over-following caused by noise.

[0030] Timing filtering: To suppress short-term noise and maintain real-time performance, a smoothed quality quantity is obtained by recursive filtering of the window quality quantity. The filter coefficients can be determined according to the sampling period and the time constant of the solidification process: a stronger smoothing can be used when the sampling period is short to suppress noise, and a weaker smoothing can be used when the sampling period is long to maintain the phase response.

[0031] Trend penalty: To expose the slow drift during the solidification phase earlier, a penalty term can be introduced to the decreasing trend of smoothing quality, forming a dynamic index of image quality. Trend penalty can be implemented based on the difference or slope between adjacent windows: when the smoothing quality is continuously decreasing, the penalty weight is increased to amplify the change; when the smoothing quality is stable or increasing, the penalty term can be reduced or reduced to zero, thereby avoiding over-response to normal noise.

[0032] Parameter examples: The preset time window can be a sliding window covering multiple consecutive frames, preferably one that covers short noise periods and is less than the main time constant that causes drift due to solidification; the weight configuration can be set according to focal length, emphasizing the outer ring and corners at the telephoto end and the midpoint and corners at the wide-angle end; the parameters of filtering and trend penalty can be determined by statistical distribution calibration of gold samples or qualified samples, so that the dynamic indicators of image quality are stable and repeatable under qualified assembly conditions.

[0033] In one embodiment, a time window is provided Inner region set With feature set The extracted original image quality features are ,in Perform dimensionless normalization on each feature dimension:

[0034] in, Statistical data obtained from gold samples or active calibration stable segments. To prevent zero constant. Concatenate them into a vector in a fixed order. and use weight vector (or weight matrix) The fusion yields the window quality value: ;right The smoothing quality value is obtained by performing recursive filtering: ;in The value is selected based on the sampling period and the time constant of the curing process. A penalty term is introduced for a continuous downward trend: ; The final image quality dynamic index is defined as: ; in This is the trend penalty coefficient. When... Located within the preset range Inside, and in continuous Within each cycle, it satisfies ; then it is assumed that the image quality is stable, in which The allowable fluctuation threshold.

[0035] The displacement measurement module is used to perform online estimation of the relative poses between the first and second module components, and between the third and second module components, during the UV curing stage, and outputs the coupled relative pose error state quantity. It also provides key state quantities for the assembly system during the UV curing stage. Dynamic changes in relative poses may occur during the curing stage, and relying solely on imaging data may be insufficient to promptly identify and locate the sources of relative pose deviations between components. Therefore, the displacement measurement module estimates the relative poses between the first and second module components and between the third and second module components online, respectively, and fuses these estimates to represent a coupled relative pose error state quantity. This state quantity serves as the input for control during the curing stage, making subsequent micro-pose compensation more targeted and controllable.

[0036] In one specific implementation, during the UV curing stage, the displacement measurement unit establishes online observation links between the zoom lens barrel assembly and the module base assembly, and between the image sensor assembly and the module base assembly, and completes the synchronization and calibration of the observation times. Specifically, the displacement measurement unit performs unified timestamp alignment on displacement, angle, or phase-type observations from different observation channels, ensuring that observations at the same time correspond one-to-one. Using the assembly reference of the module base assembly as a coordinate reference, the unit performs geometric calibration and coordinate registration on each observation channel, mapping the observations to a relative pose expression within the same coordinate frame. For phase-type observations, the displacement measurement unit performs phase unwrapping processing in the sampling sequence to avoid introducing jump errors due to small continuous movements crossing the phase period during curing. Simultaneously, the displacement measurement unit introduces rigid body motion constraints, treating the zoom lens barrel assembly and the image sensor assembly as rigid bodies, limiting their translation and rotation to satisfy rigid body geometric consistency, thereby maintaining the physical feasibility of relative pose solution even in the presence of observation noise. For example, when UV irradiation causes the adhesive layer to shrink and the image sensor assembly to tilt slowly, time synchronization ensures that the tilt change and displacement change are captured at the same time, and calibration and registration ensure that the change is correctly interpreted as an attitude change relative to the module base assembly rather than a sensor mounting bias. After preprocessing, the displacement measurement unit establishes a state-space model by combining multi-source fusion nonlinear state estimation and outputs coupled relative pose error state variables. The coupled relative pose error state variables include the joint state of six-degree-of-freedom relative pose deviations and their dynamic terms. Specifically, the state vector includes at least the three-dimensional translational and rotational deviations of the zoom lens assembly relative to the module base assembly, and the three-dimensional translational and rotational deviations of the image sensor assembly relative to the module base assembly. It further includes dynamic terms used to describe the slow drift and transient disturbances during the solidification stage, such as the first-order rate of change of the relative pose deviation or the equivalent disturbance term. The observation equation maps the registration observations obtained in step one to constraints on the above state vectors and incorporates the phase-unwrapped observations into the update process in a continuous form. The state transition equation introduces the slowly varying prior of the solidification process based on the rigid body motion constraints, so that the estimation can respond to short-term disturbances and track the gradual drift caused by solidification. The displacement measurement unit iteratively updates the joint state in each sampling period through nonlinear state estimation, so that the output coupled relative pose error state quantity simultaneously reflects the deviation and coupling relationship of the two relative links. For example, when the zoom lens barrel assembly has a slight lateral eccentricity and the image sensor assembly has a small angle tilt, multi-source fusion can use the differences in sensitivity of different observation channels to translation and rotation to separate and reconstruct their respective six-degree-of-freedom relative pose deviations. At the same time, it uses dynamic terms to characterize the evolution rate of the deviation with the solidification time, providing a predictable and convergent state input for subsequent micro-pose compensation.

[0037] In one embodiment, the coupled relative pose error state variables can be estimated online using the following state-space model: Coordinates and Calibration: A reference coordinate system is established based on the assembly datum of the second module component. The displacement, angle, and phase observation channels are mapped to the same reference coordinate system through geometric calibration and coordinate registration. For phase observations, phase unwrapping is performed first, and installation fixed offset terms are removed to ensure that the observations are continuous over time.

[0038] State Definition: The three-dimensional translational and rotational deviations of the first module component relative to the second module component, and the three-dimensional translational and rotational deviations of the third module component relative to the second module component, are defined as pose components of the coupled state. A dynamic term is introduced to describe the slow drift and transient disturbances during the solidification phase; this dynamic term can be the first-order rate of change of the pose deviation or an equivalent disturbance term. This constitutes a joint state vector, which simultaneously characterizes the deviations of the two relative links and their evolutionary trends over time.

[0039] State transition: Within the discrete sampling period, the state transition can adopt the prior model of "slowly changing drift priority": the pose deviation is extrapolated from the pose deviation of the previous period and the dynamic term. The dynamic term changes slowly in adjacent periods and is superimposed with process noise to characterize the uncertain disturbances caused by adhesive shrinkage, thermal effects and stress release.

[0040] Observation Model: The observation vector consists of registered displacement, angle, and phase observations. The observation equation maps these observations to constraints on the joint state vector. For multi-point displacement observations, rigid body motion constraints can be used to introduce multi-point displacement consistency into the construction of observation residuals, thereby improving the observability of translation and rotation and suppressing noise amplification.

[0041] Online estimation: Nonlinear state estimation can be used to recursively update the joint state in each sampling period, including prediction and correction steps: the prediction step extrapolates the prior state and covariance based on the state transition; the correction step updates the posterior state and covariance based on the observation residuals. Through multi-source fusion, the differences in the sensitivity of different observation channels to translation and rotation are utilized to achieve separate estimation of the six-degree-of-freedom deviation of the two links, and the drift trend is output by the dynamic term.

[0042] Sampling and initialization: The sampling period can be consistent with the window update cycle of the imaging evaluation or its integer frequency division; initialization can be performed before the active calibration ends and the solidification begins, based on the initial state and noise statistics estimated by the static sampling segment, so as to ensure the estimation convergence speed and stability during the solidification stage.

[0043] The collaborative control module connects to the dual-motion platform module and the ultraviolet curing light source. The collaborative control module is used to achieve unified coordination of "perception-decision-execution." On one hand, it connects to the dual-motion platform module, translating control quantities into pose-type commands; on the other hand, it connects to the ultraviolet curing light source, translating control quantities into irradiation energy adjustment commands. Through this connection, the collaborative control module can invoke different inputs (image quality dynamic indicators, coupled relative pose error state quantities) at different stages, and drive pose adjustment and curing process adjustment respectively, achieving coordination, consistency, and continuity between the two stages.

[0044] During the active calibration phase, the pose calibration command of the dual-motion platform module is output based on the dynamic image quality index. The control closed loop of the active calibration phase uses the dynamic image quality index as a direct evaluation basis. The collaborative control module generates a pose calibration command based on the trend and current level of the index and sends it to the dual-motion platform module. After the dual-motion platform module executes the pose calibration command, it gradually adjusts the pose of the first and third module components relative to the second module component to a state that can obtain the target imaging quality, thereby completing the target alignment and optimization of the active calibration phase.

[0045] During the UV curing stage, the dual-motion platform module outputs a micro-pose compensation command based on the image quality dynamic index and the coupled relative pose error state quantity. The dynamic compensation closed loop during the curing stage emphasizes "micro," "real-time," and "continuous" pose correction. The collaborative control module simultaneously utilizes the image quality dynamic index (reflecting changes in imaging quality) and the coupled relative pose error state quantity (reflecting relative pose deviation) to generate a micro-pose compensation command and output it to the dual-motion platform module. Based on this, the dual-motion platform module performs small-amplitude pose compensation on the first and third module components, suppressing image quality fluctuations during curing and reducing the risk of image quality deviation after curing.

[0046] During the UV curing stage, an irradiation energy adjustment command is output to the UV curing light source. The "optical-mechanical collaborative" control approach in the curing stage is as follows: In addition to pose compensation, the energy input of the curing process itself also affects the stability of the curing process and the final state. Therefore, the collaborative control module outputs an irradiation energy adjustment command to the UV curing light source to coordinate and adjust the curing process. By simultaneously implementing energy adjustment and micro-pose compensation, the assembly state during the curing stage becomes more controllable, which helps to maintain the dynamic indicators of image quality within a preset range when curing is complete.

[0047] When the coupled relative pose error state quantity meets the preset convergence criterion or the preset solidification criterion, the output of micro-pose compensation commands is stopped. The stopping conditions and engineering boundaries for micro-pose compensation during the solidification phase are as follows: when the coupled relative pose error state quantity reaches the convergence requirement, it indicates that the relative pose deviation has entered an acceptable range; or when the preset solidification criterion is met, it indicates that the solidification process has reached the predetermined stage / end point, at which point continued compensation may no longer be necessary or economical. Therefore, the collaborative control module stops outputting micro-pose compensation commands to ensure that the control process has clear termination conditions and repeatable process criteria.

[0048] In one specific implementation, during the active calibration phase, the collaborative control module receives the image quality dynamic index output by the imaging evaluation module and determines the image quality dynamic index as the target value. The target value can be set as the optimal index achievable under the current operating conditions or a pre-calibrated qualified upper limit index. The collaborative control module calculates the deviation between the image quality dynamic index and the target value and uses this deviation as a driving quantity to generate a pose calibration command. The pose calibration command is sent to the dual-motion platform module to drive the dual-motion platform module to adjust the pose of the first module component and / or the third module component, so that the image quality dynamic index approaches the target value. To ensure that the calibration process is both efficient and repeatable, the pose calibration command can adopt a unified action sequence of coarse-to-fine. For example, first, a three-way translation is used to quickly converge the index loss caused by optical axis eccentricity to the adjustable range, then a three-way rotation is used to eliminate off-axis degradation caused by tilt angle, and finally, a small translation is performed along the optical axis to compensate for the image plane position. After each adjustment, the image quality dynamic index is reacquired and the deviation is updated, thus forming a closed-loop iteration. For example, when the indicators of the four corner sub-regions at the telephoto end decrease while the indicators of the central region of the field of view remain basically unchanged, the collaborative control module can prioritize generating small pose calibration commands around the pitch and roll axes to reduce the off-axis inconsistency caused by astigmatism and field curvature. Then, after the indicators of the outer ring region reach the neighborhood of the target value, it switches to pose calibration commands with smaller steps to avoid over-adjustment leading to a decrease in contrast in the central region. During the UV curing stage, the collaborative control module continuously receives the dynamic index of image quality and the coupled relative pose error state quantity output by the displacement measurement module. Based on the deviation of the dynamic index of image quality from the target value and the coupled relative pose error state quantity, the module performs model prediction extrapolation to obtain the predicted coupled pose error state quantity. After obtaining the predicted coupled pose error state quantity, the collaborative control module generates pose prediction compensation trajectories for the first module component and the third module component, and generates and outputs micro-pose compensation commands to the dual-motion platform module based on the pose prediction compensation trajectories. To enable feedforward / predictive capabilities in the compensation process during curing, model prediction extrapolation can process the gradual drift and transient disturbances caused by curing in layers. For example, the dynamic term in the coupled relative pose error state quantity can be extrapolated in a short time to predict the drift direction of the next control cycle. This predicted quantity is then superimposed on the feedback correction quantity obtained from the deviation of the image quality dynamic index. This allows the micro-pose compensation command to suppress instantaneous fluctuations in the index and to offset the trend shift caused by adhesive shrinkage or thermal effects in advance. At the same time, the pose prediction compensation trajectory can include constraint boundaries for displacement, rotation, and acceleration, so that the compensation action remains continuous without introducing vibration. For example, when the coupled relative pose error state quantity shows that the third module component has a slowly increasing rotational deviation around the horizontal axis relative to the second module component, and the image quality dynamic index shows a continuous downward trend in the edge region, the collaborative control module can extrapolate to obtain the predicted coupled pose error state quantity and generate a pose prediction compensation trajectory with reverse micro-rotation as the main component and supplemented by micro-translation along the optical axis direction, so that the modulation transfer function of the four corner sub-regions is suppressed in advance, thereby pulling the image quality dynamic index back to the target value neighborhood before the solidification is finished; During the UV curing stage, the collaborative control module outputs an irradiation energy adjustment command to the UV curing light source. This command adjusts at least one of the following: irradiance, irradiation duration, and duty cycle, ensuring that the dynamic image quality indicators remain within a preset range upon completion of curing. The irradiation energy adjustment command and the micro-pose compensation command can be updated collaboratively using the same sampling cycle to achieve joint stability of illumination input and pose state. To improve the process window and image quality stability, the irradiation energy adjustment command can employ a segmented strategy. For example, in the initial stage, pre-curing is performed with lower irradiance and a lower duty cycle to reduce the rate of increase in coupled relative pose error caused by rapid contraction. In the intermediate stage, the irradiance or duty cycle is gradually increased to ensure sufficient curing. In the final stage, the irradiance is reduced and the irradiation duration is extended to reduce overshoot. Simultaneously, the dynamic image quality indicators are continuously monitored to ensure they fall within the preset range and remain stable. When the coupled relative pose error state quantity meets the preset convergence criterion or the preset curing criterion, the collaborative control module stops outputting the micro-pose compensation command to avoid introducing unnecessary disturbances at the end of curing. For example, the preset convergence criterion can be set to the norm of the coupled relative pose error state quantity being lower than the threshold within a certain number of consecutive sampling periods and the fluctuation of the image quality dynamic index within the preset range not exceeding the allowable range. The preset solidification criterion can be set to the cumulative irradiation energy reaching the process requirements and the image quality dynamic index remaining within the preset range. Once one of these conditions is met, the collaborative control module terminates the micro-pose compensation command and maintains the irradiation energy adjustment command to complete the final solidification, so that the image quality dynamic index is in an acceptable state when solidification is completed.

[0049] In one embodiment, the state vector coupling the relative pose error state variables can be taken as:

[0050] Among them, superscript Indicates the first module component relative to the second module component, superscript This indicates the third module component relative to the second module component; This is a three-dimensional translational deviation. For three-way rotational deviation, This represents the first-order rate of change. The state transition uses a first-order uniform drift model: ;in Here is the state transition matrix. This represents the process noise term. Observation vector. It consists of registered displacement / angle / phase observations, and the observation model is as follows: ; in It is a nonlinear mapping consisting of rigid body constraints and calibration matrices. For the observation noise term, recursive estimation can be performed using extended Kalman filtering (EKF) or unscented Kalman filtering (UKF). Model prediction extrapolation can be performed using... ; The predicted coupled pose error state variables are obtained, where For a moment The posterior estimate. The infinitesimal pose compensation can be synthesized from the feedback term and the feedforward term: ; in The target value for the dynamic index of image quality. This represents the gain matrix / coefficients. And for... After applying single-cycle displacement / rotation / acceleration limiting and smoothing constraints, the output is a micro-pose compensation command to ensure that the compensation trajectory is continuous and executable.

[0051] In one embodiment, the micro-pose compensation command and the irradiation energy adjustment command during the curing stage can be generated according to the following cooperative strategy: Command synthesis: Micro-pose compensation is synthesized from feedback and feedforward terms. The feedback term is generated based on the deviation of the image quality dynamic index from the target value and is used to suppress image quality deviation; the feedforward term is extrapolated from the dynamic term of the coupled relative pose error state quantity and is used to offset trend drift in advance. The synthesized compensation amount is then subjected to amplitude limiting and smoothing to form a compensation command to avoid introducing vibration.

[0052] Limiting and smoothing: To ensure mechanical stability, the compensation command can be set with a single-cycle displacement limit, a single-cycle rotation angle limit, and acceleration / jerk constraints, making the compensation trajectory continuous and executable; when the estimated noise increases or the image quality index fluctuates more, the compensation step size can be automatically reduced and the smoothing intensity can be increased to suppress the risk of overcompensation.

[0053] Timing coordination: Micro-pose compensation and irradiation energy adjustment can be updated synchronously using the same sampling time. When the curing illumination interferes with imaging, a time-sharing strategy can also be adopted: images are acquired and indicators and state quantities are updated in non-irradiation time slots, while the platform is kept stationary or only a smaller amount of compensation is performed in irradiation time slots, thereby balancing the reliability of imaging evaluation and the continuity of curing.

[0054] Energy regulation strategy: Irradiation energy regulation can be carried out in stages: in the pre-curing stage, the irradiance or duty cycle is reduced to slow down the pose change caused by rapid contraction; in the intermediate stage, the irradiance or duty cycle is increased to ensure sufficient curing; in the final stage, the irradiance is reduced and the irradiation time is appropriately extended to reduce overshoot and thermal disturbance. The segment switching can be triggered by the stability of the image quality dynamic index, the drift rate of the coupled relative pose error state quantity, or the cumulative irradiation energy.

[0055] Stop compensation protection: When entering the curing end and meeting the preset convergence criterion or preset curing criterion, stop outputting micro pose compensation commands and keep the platform stationary to reduce the impact of end mechanical disturbance and rebound on the final locked state, while continuing or maintaining the necessary irradiation energy to complete the final curing.

[0056] In one embodiment, the target value, preset range, threshold, and criterion mentioned in the specification can be determined in the following operable manner to ensure that the present invention can be implemented by those skilled in the art: Target value and preset range determination: Select gold samples or verified zoom lens module samples, and collect imaging data under specified target, focal length, and exposure conditions to calculate the statistical distribution of dynamic image quality indicators. Set the target value as the representative value of this distribution, and set the preset range as the allowable fluctuation range around the target value; the allowable fluctuation range can be determined based on the dispersion of the good product distribution, the product image quality acceptance threshold, and the production line environmental noise, so that the indicators of qualified samples fall within the preset range and have sufficient robustness when the solidification is completed.

[0057] Pose threshold determination: Establishing a sensitivity relationship between pose changes and changes in dynamic image quality indicators can be achieved by applying small, known perturbations to each degree of freedom after active calibration and observing the changes in indicators. Based on this sensitivity relationship, the "allowable decrease in indicators" is used to deduce the "allowable pose deviation threshold," thus providing a basis for setting the threshold for coupled relative pose error state quantities.

[0058] Preset convergence criteria are determined as follows: The preset convergence criteria can be composed of "the pose error enters the allowable range and remains stable", that is, the coupled relative pose error state quantity remains below the pose threshold for multiple consecutive sampling periods, while the dynamic index of image quality is within the preset range and its fluctuation amplitude does not exceed the allowable amplitude; the number of consecutive periods can be selected in association with the main time constant of the curing process, preferably so that it can cover the stress release or thermal recovery process at the end of curing.

[0059] Preset curing criteria are determined as follows: The preset curing criteria can be composed of "cumulative irradiation energy reaches the process requirements" and "image quality indicators remain stable". The process requirements for cumulative irradiation energy can be derived from the curing curve of the adhesive material supplier or the curing strength threshold determined by the spline test; in order to take into account batch differences, a process margin can be set for the cumulative energy requirements, and the energy input in the final stage can be jointly constrained with the image quality stability.

[0060] Parameter calibration process: The above thresholds and ranges can be obtained through gold sample calibration during the production line introduction stage, and periodically reviewed and updated through random sampling during the mass production stage; when the batch of adhesive material is changed, the light source irradiance is calibrated, or the assembly tolerance window changes, sensitivity calibration and statistical calibration can be re-executed to ensure the traceability and consistency of thresholds and criteria.

[0061] Delete or replace the catch-all sentence: To avoid the risk of insufficient disclosure due to simply stating "set by those skilled in the art", the preset parameters involved in the specification shall at least give the determination process and influencing factors, so that those skilled in the art can complete the implementation and parameter tuning accordingly.

[0062] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A zoom lens module active calibration and multi-degree-of-freedom coupling debugging device, characterized in that, include: Positioning and clamping module, dual-motion platform module, imaging evaluation module, displacement measurement module, and collaborative control module; The positioning and clamping module is used to fix the second module component and provide an assembly reference. The dual-motion platform module is used to support the first module component and the third module component respectively, and to achieve multi-degree-of-freedom pose adjustment of three-way translation and three-way rotation relative to the second module component; The imaging evaluation module is used to acquire imaging data output by the third module component during the active calibration phase and the ultraviolet curing phase, and output dynamic indicators of image quality. The displacement measurement module is used to estimate the relative pose between the first module component and the second module component, and between the third module component and the second module component, during the UV curing stage, and output the coupled relative pose error state quantity. The collaborative control module connects to the dual-motion platform module and the ultraviolet curing light source. The collaborative control module is used for: During the active calibration phase, pose calibration commands for the dual-motion platform module are output based on dynamic image quality indicators. During the UV curing stage, the dual-motion platform module outputs a micro-pose compensation command based on the dynamic index of image quality and the coupled relative pose error state quantity. During the ultraviolet curing stage, an irradiation energy adjustment command is output to the ultraviolet curing light source; When the coupled relative pose error state quantity meets the preset convergence criterion or the preset solidification criterion, the output of the micro pose compensation command stops.

2. The zoom lens module active calibration and multi-degree-of-freedom coupling debugging device according to claim 1, characterized in that, The first module component is a zoom lens barrel assembly, the second module component is a module base assembly, and the third module component is an image sensor assembly.

3. The zoom lens module active calibration and multi-degree-of-freedom coupling debugging device according to claim 1, characterized in that, The positioning and clamping module includes a reference platform, at least three non-collinear positioning reference elements, and a clamping mechanism. The positioning reference elements include at least two of the following: a reference plane, a positioning pin, and a stop surface. The second module component has a reference surface and / or a reference hole that mates with the positioning reference elements. After mating, the clamping mechanism presses or adsorbs the second module component onto the reference platform to establish an assembly reference.

4. The zoom lens module active calibration and multi-degree-of-freedom coupling debugging device according to claim 1, characterized in that, The imaging evaluation module extracts and normalizes multi-region image quality features based on imaging data within a preset time window. The normalized multi-region image quality features constitute a feature vector. The feature vector is fused by weight matrix mapping and combined with temporal filtering and trend penalty processing to form a dynamic index of image quality.

5. The zoom lens module active calibration and multi-degree-of-freedom coupling debugging device according to claim 4, characterized in that, The multiple regions include the central region of the field of view, at least one inner ring region, at least one outer ring region, and edge regions. The edge regions include at least one type of four-corner sub-regions and edge midpoint sub-regions. The weights of each region are configured according to the field of view position and image quality sensitivity, and are used for weight matrix mapping and fusion.

6. The zoom lens module active calibration and multi-degree-of-freedom coupling debugging device according to claim 4, characterized in that, Image quality features include one or more of the following: modulation transfer function, spatial frequency response, contrast, astigmatism measure, and field curvature measure.

7. The zoom lens module active calibration and multi-degree-of-freedom coupling debugging device according to claim 1, characterized in that, Online estimation is based on observation time synchronization and calibration registration, and establishes a state space model by combining phase unwrapping, rigid body motion constraints and multi-source fusion nonlinear state estimation; outputs coupled relative pose error state quantity, which includes the joint state of six-degree-of-freedom relative pose deviation and its dynamic terms.

8. The zoom lens module active calibration and multi-degree-of-freedom coupling debugging device according to claim 1, characterized in that, The collaborative control module is used for: During the active calibration phase, the image quality dynamic index is determined as the target value, and a pose calibration command is generated based on the deviation between the image quality dynamic index and the target value to drive the dual-motion platform module to adjust the pose of the first module component and / or the third module component so that the image quality dynamic index approaches the target value. During the UV curing stage, the model prediction extrapolates the predicted coupled pose error state quantity based on the deviation of the image quality dynamic index relative to the target value and the coupled relative pose error state quantity. Then, pose prediction compensation trajectories for the first module component and the third module component are generated, and micro pose compensation commands are generated and output based on the pose prediction compensation trajectories.

9. The zoom lens module active calibration and multi-degree-of-freedom coupling debugging device according to claim 1, characterized in that, The irradiation energy adjustment command is used to adjust at least one of the irradiance, irradiation duration, and duty cycle to keep the dynamic indicators of image quality within a preset range when curing is complete.

10. A zoom lens module active calibration and multi-degree-of-freedom coupling debugging system, characterized in that, It is applied to the active calibration and multi-degree-of-freedom coupling debugging device for zoom lens modules as described in any one of claims 1-9.

Images