Microlens and pixel array alignment method, device, system and storage medium

Abstract

This invention relates to a method, system, apparatus, and storage medium for aligning microlenses and pixel arrays. The method includes: acquiring a first marked image of a pixel array and a second marked image of a microlens array in a CMOS sensor chip; calculating the offset between the first and second marked images to obtain a target offset; when the target offset is detected to be inconsistent with a preset alignment accuracy threshold, performing position compensation on the target offset until the target offset meets the preset alignment accuracy threshold; calculating the relative positional drift between the microlens array and the pixel array based on the response signal acquired by the photoelectric detection array; dynamically aligning and maintaining the microlens array and the pixel array based on the relative positional drift; and generating an alignment accuracy report. This invention, by acquiring marked images of a CMOS sensor chip and performing offset calculations, can accurately identify misalignment errors between the two.

Description

Technical Field

[0001] This invention relates to the technical field of microlenses and pixel arrays, and particularly to a method, apparatus, system, and storage medium for aligning microlenses and pixel arrays. Background Technology

[0002] High-precision alignment of microlenses and pixel arrays is a critical manufacturing step for improving the performance of CMOS sensors, directly impacting the final quality of high dynamic range imaging modules. Current methods often rely on single static positioning and bonding, which struggles to overcome micron-level positional drift caused by assembly stress, material deformation, and curing shrinkage. This results in decreased optical path coupling efficiency between lenses and pixels and compromised imaging uniformity. Especially under wide temperature, vibration, or long-term operating conditions, existing technologies lack real-time monitoring and dynamic compensation capabilities, failing to suppress drift in a closed-loop manner during bonding and curing. This makes it difficult to maintain stable alignment accuracy, severely limiting the reliability and consistency of sensors in complex scenarios such as backlighting and low illumination. Summary of the Invention

[0003] The main objective of this invention is to provide a method, apparatus, system, and storage medium for aligning microlenses and pixel arrays. By acquiring marked images of pixel arrays and microlens arrays in a CMOS sensor chip and performing offset calculations, the misalignment error between the two can be accurately identified.

[0004] To achieve the above objectives, the present invention provides a method for aligning a microlens with a pixel array, comprising: The first marker image of the pixel array and the second marker image of the microlens array in the CMOS sensor chip are acquired, and the offset of the first marker image and the second marker image is calculated to obtain the target offset. When the target offset is detected to be not satisfied with the preset alignment accuracy threshold, position compensation is performed on the target offset until the target offset satisfies the preset alignment accuracy threshold. When the target offset is detected to meet the preset alignment accuracy threshold, the photoelectric detection array around the pixel array is activated, and the relative position drift between the microlens array and the pixel array is calculated based on the response signal collected by the photoelectric detection array. The microlens array and the pixel array are dynamically aligned and maintained based on the relative position drift, and an alignment accuracy report is generated.

[0005] Further, the acquisition of a first marker image of the pixel array and a second marker image of the microlens array in the CMOS sensor chip, and the calculation of the offset between the first marker image and the second marker image to obtain the target offset, includes: The center coordinates of the markers are extracted from the first marker image and the second marker image respectively, resulting in a first coordinate list and a second coordinate list; According to the preset marker layout, the first coordinate list and the second coordinate list are paired to form a set of matching point pairs; Calculate the coordinate difference between each pair of coordinate points in the horizontal and vertical directions in the matching point pair set to obtain the original offset set; Each offset in the original offset set is compared with a preset tolerance threshold, and the offsets exceeding the tolerance threshold are removed to obtain the effective offset set; Calculate the horizontal average of all horizontal offsets and the vertical average of all vertical offsets in the effective offset set, and integrate the horizontal average and the vertical average to obtain the target offset.

[0006] Further, the step of performing position compensation on the target offset when it is detected that the target offset does not meet the preset alignment accuracy threshold, until the target offset meets the preset alignment accuracy threshold, includes: The target offset is compared with the preset alignment accuracy threshold to determine whether the absolute value of the target offset is greater than the preset alignment accuracy threshold. When the absolute value of the target offset is greater than the preset alignment accuracy threshold, the target offset is subjected to direction compensation calculation to obtain position compensation information; The microlens array is driven to move in the opposite direction of the target offset based on the position compensation information; After the microlens array moves, the second marker image is reacquired, and the offset between the second marker image and the first marker image is calculated to update the target offset until the absolute value of the updated target offset is less than the preset alignment accuracy threshold.

[0007] Further, when the target offset is detected to meet the preset alignment accuracy threshold, the photoelectric detection array around the pixel array is activated, and the relative positional drift between the microlens array and the pixel array is calculated based on the response signal collected by the photoelectric detection array, including: The photoelectric detection array is activated to enter the working state and receives the raw signals output by each detection unit in the photoelectric detection array; The original signal is filtered according to a preset background noise threshold to obtain valid response signals, and the physical coordinates of each valid response signal are obtained. The signal deviation value is obtained by calculating the unit difference between each effective response signal and the preset reference signal. Based on the signal deviation value and the physical coordinates, the spatial centroid is calculated to obtain the signal centroid coordinates; The relative position drift is obtained by calculating the drift position of the signal centroid coordinates and aligning them with the preset reference coordinates.

[0008] Further, the step of filtering the original signal according to a preset background noise threshold to obtain valid response signals, and acquiring the physical coordinates of each valid response signal, includes: The intensity of each of the original signals is detected to obtain the signal intensity value; Each of the signal strength values ​​is compared with the preset background noise threshold; If the signal strength value is greater than the preset background noise threshold, then the corresponding original signal is marked as the valid response signal; The physical coordinates are obtained by performing coordinate matching based on the effective response signal and the preset unit coordinate relationship.

[0009] Further, the step of calculating the spatial centroid based on the signal deviation value and the physical coordinates to obtain the signal centroid coordinates includes: The physical coordinates and signal deviation values ​​associated with each valid response signal are read sequentially. Multiply the horizontal coordinate of each physical coordinate by the corresponding signal deviation value, and sum the resulting horizontal weighted values ​​to obtain a horizontal weighted sum; Multiply the ordinate of each physical coordinate by the corresponding signal deviation value, and sum the resulting longitudinal weighted values ​​to obtain a longitudinal weighted sum; The total deviation is obtained by summing all the signal deviation values. Divide the horizontal weighted sum by the total deviation sum to obtain the horizontal coordinate of the signal centroid; Divide the weighted sum of the longitudinal axes by the total sum of deviations to obtain the longitudinal coordinate of the signal centroid; By integrating the horizontal coordinate of the signal centroid and the vertical coordinate of the signal centroid, the signal centroid coordinates are obtained.

[0010] Further, the step of dynamically aligning and maintaining the microlens array and the pixel array based on the relative position drift, and generating an alignment accuracy report, includes: The relative position drift is analyzed to obtain the drift direction and drift amplitude, and the drift amplitude is compared with a preset drift threshold. When the drift amplitude exceeds the drift threshold, a preset compensation mapping relationship is queried based on the drift direction to obtain the position compensation direction. The position compensation direction is then used to construct a compensation control signal by compensating for the drift amplitude. The microlens array and the pixel array are aligned and compensated according to the compensation control signal, and the new relative position drift is re-acquired for detection. When the drift amplitude is less than the drift threshold, the current alignment state is recorded, and an alignment accuracy report is generated based on the previous compensation records.

[0011] The present invention also provides a microlens and pixel array alignment device, applied to the microlens and pixel array alignment method described in any one of the above claims, comprising: The acquisition module is used to acquire a first marker image of the pixel array and a second marker image of the microlens array in the CMOS sensor chip, and to perform offset calculation on the first marker image and the second marker image to obtain the target offset. The analysis module is used to perform position compensation on the target offset when it is detected that the target offset does not meet the preset alignment accuracy threshold, until the target offset meets the preset alignment accuracy threshold. The association module is used to activate the photoelectric detection array around the pixel array when the target offset is detected to meet the preset alignment accuracy threshold, and calculate the relative position drift between the microlens array and the pixel array based on the response signal collected by the photoelectric detection array. The processing module is used to dynamically align and maintain the microlens array and the pixel array based on the relative position drift amount, and generate an alignment accuracy report.

[0012] The present invention also provides a microlens and pixel array alignment system, comprising: Memory, used to store programs; A processor is configured to execute the program to implement the various steps of the microlens and pixel array alignment method described in any of the preceding claims.

[0013] The present invention also provides a storage medium storing computer instructions for causing a computer to perform any of the methods described above.

[0014] The microlens and pixel array alignment method, apparatus, system, and storage medium provided by this invention have the following beneficial effects: By acquiring marked images of the pixel array and microlens array in the CMOS sensor chip and performing offset calculations, misalignment errors between them can be accurately identified, providing a reliable data foundation for high-precision alignment and effectively improving the accuracy of initial positioning. Iterative position compensation is performed when the offset does not meet the accuracy threshold until the preset requirements are met, gradually correcting deviations caused by assembly and environmental factors, enhancing the stability and repeatability of the alignment process, and reducing the need for manual intervention. By activating a photoelectric detection array to monitor drift in real time and dynamically maintaining alignment during the bonding stage, drift caused by material deformation and curing shrinkage can be suppressed, ensuring the reliability of alignment accuracy during long-term operation, thereby improving imaging uniformity and product performance consistency. Attached Figure Description

[0015] Figure 1 This is a flowchart of the microlens and pixel array alignment method provided by the present invention; Figure 2 This is a structural diagram of the microlens and pixel array alignment device provided by the present invention; Figure 3 This is a structural diagram of the microlens and pixel array alignment system provided by the present invention.

[0016] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments.

[0019] Reference Figure 1 As shown, the present invention provides a method for aligning a microlens with a pixel array, comprising: Step S1: Acquire the first marker image of the pixel array and the second marker image of the microlens array in the CMOS sensor chip, and perform offset calculation on the first marker image and the second marker image to obtain the target offset; Specifically, optical alignment marks with specific geometric shapes are fabricated in the peripheral region of the pixel array of the CMOS sensor chip and at corresponding positions of the microlens array. Digital images of the two sets of marks are simultaneously acquired under uniform illumination using a high-resolution optical acquisition system. The acquired first and second mark images are preprocessed, including grayscale conversion, noise filtering, and edge enhancement. Then, a contour detection algorithm is used to extract the closed boundaries of each mark and calculate its pixel-level geometric center coordinates. Based on a preset spatial distribution rule for the marks, the center coordinates of the marks in the two sets of images are matched one-to-one to establish a set of coordinate point pairs. For each matched point pair, the coordinate difference in the X and Y axes of the image coordinate system is calculated to form an original set of offset vectors. A tolerance threshold based on process experience is set, and each offset vector in this set is filtered to remove abnormal offset data caused by mark defects or image noise, retaining only the effective offsets that conform to statistical laws. Finally, the arithmetic mean of all X-axis components in the effective offset set is calculated as the horizontal target offset component, and the arithmetic mean of the Y-axis components is calculated as the vertical target offset component.

[0020] Step S2: When the target offset is detected to be not satisfied with the preset alignment accuracy threshold, position compensation is performed on the target offset until the target offset satisfies the preset alignment accuracy threshold; Specifically, the horizontal and vertical components of the target offset are compared with pre-set alignment accuracy thresholds. If the absolute value of either component is greater than the corresponding accuracy threshold, the alignment requirement is not met, and the compensation process begins. Based on the direction of the misalignment and the magnitude of the offset, a control command containing the compensation direction and amount is generated. This command is sent to the displacement mechanism driving the microlens array, which drives the microlens array to translate in the opposite direction of the target offset. After a single displacement compensation, the image acquisition and offset calculation process of step S1 is re-executed to obtain the updated target offset. The updated target offset is compared again with the same accuracy threshold. If the requirement is still not met, the cycle of generating the command, driving the displacement, recalculating, and comparing is repeated. This iterative compensation mechanism continues until the absolute values ​​of the latest calculated target offset components in both directions are not greater than the corresponding preset accuracy thresholds. At this point, the microlens array and pixel array are determined to have reached a preliminary precision alignment state, and the compensation cycle ends.

[0021] Step S3: When the target offset is detected to meet the preset alignment accuracy threshold, the photoelectric detection array around the pixel array is activated, and the relative position drift between the microlens array and the pixel array is calculated based on the response signal collected by the photoelectric detection array. Specifically, once the target offset meets a preset accuracy threshold, a photoelectric detection array integrated into the periphery of the pixel array is activated. This array consists of multiple regularly distributed photosensitive detection units, each of which outputs a current signal related to the incident light intensity under uniform light source illumination of a specific wavelength. The raw current signals output by all detection units are collected at a fixed sampling period to form a raw signal sequence. A background noise threshold is set based on a pre-calibrated background noise level, and the raw signal sequence is filtered point by point, retaining only valid response signals with amplitudes exceeding the threshold, and recording the corresponding physical coordinates of the detection units. The real-time amplitude of each valid response signal is compared with the pre-stored reference signal amplitude acquired under ideal alignment conditions to calculate the signal deviation value of each unit. Based on the physical coordinates of all valid units and their corresponding signal deviation values, spatial weighted centering calculation is performed. The X-coordinate of each unit is multiplied by its signal deviation value and summed to obtain the X-direction weighted sum; the Y-coordinate undergoes the same processing to obtain the Y-direction weighted sum. All signal deviation values ​​are then summed as the total weight. The X-coordinate of the signal centroid is obtained by weighting the sum in the X direction and dividing by the total weight, and the Y-coordinate is obtained by weighting the sum in the Y direction and dividing by the total weight. Finally, the calculated signal centroid coordinates are compared with the pre-stored theoretical alignment center coordinates, and the difference between the two coordinates in the X and Y directions is defined as the relative positional drift between the microlens array and the pixel array.

[0022] Step S4: Dynamically align and maintain the microlens array and the pixel array based on the relative position drift, and generate an alignment accuracy report.

[0023] Specifically, during the physical bonding stage between the microlens array and the CMOS sensor chip using optical adhesive, closed-loop control is continuously executed based on the real-time relative position drift. The horizontal and vertical components of this drift are compared in real-time with a set of more stringent dynamic maintenance accuracy thresholds. When the component in either direction exceeds its corresponding maintenance threshold, a corresponding fine-tuning compensation command is generated based on the direction and magnitude of that component. This command is converted into a drive signal, controlling the precision micro-motion mechanism connected to the microlens array to perform sub-micron-level reverse displacement to actively counteract the detected position drift. Throughout the bonding stage (including adhesive flow, initial curing, and final curing), the process of "drift monitoring - threshold judgment - command generation - pose fine-tuning" is executed cyclically at millisecond intervals. When the drift over multiple consecutive cycles remains stable within the maintenance threshold, the alignment is considered to have reached final stability. Subsequently, key data from the entire process is automatically summarized and recorded, including the initial target offset sequence, details of each compensation command, the drift history for all monitoring cycles, and the final stable coordinates. This data is structured and packaged to generate an alignment accuracy report that includes an overview of the alignment process and final performance metrics.

[0024] The microlens and pixel array alignment method provided by this invention can accurately identify misalignment errors between the pixel array and microlens array in a CMOS sensor chip by acquiring marked images and calculating offsets. This provides a reliable data foundation for high-precision alignment and effectively improves the accuracy of initial positioning. By iteratively compensating for positional discrepancies when the offset does not meet the accuracy threshold until a preset requirement is met, deviations caused by assembly and environmental factors can be gradually corrected, enhancing the stability and repeatability of the alignment process and reducing the need for manual intervention. By activating a photoelectric detection array to monitor drift in real time and dynamically maintaining alignment during the bonding stage, drift caused by material deformation and curing shrinkage can be suppressed, ensuring the reliability of alignment accuracy during long-term operation, thereby improving imaging uniformity and product performance consistency.

[0025] In one embodiment, acquiring a first marker image of the pixel array and a second marker image of the microlens array in a CMOS sensor chip, and performing offset calculations on the first marker image and the second marker image to obtain a target offset, includes: The center coordinates of the markers are extracted from the first marker image and the second marker image respectively, resulting in a first coordinate list and a second coordinate list; According to the preset marker layout, the first coordinate list and the second coordinate list are paired to form a set of matching point pairs; Among them, the preset marker layout refers to the strict spatial correspondence rules between two sets of markers that have been determined in the chip and microlens array design stage. These rules define, for example, the number of markers, their relative positions (such as matrix arrangement, location at the four corners, etc.), and the unique ID features (such as a specific shape sequence) that exist, in order to unambiguously establish the mapping between points in the two coordinate lists.

[0026] Specifically, based on the marker layout rules pre-stored in the system, the obtained first coordinate list and second coordinate list are associated and matched according to the topological structure of the markers. For example, if the markers are arranged in a regular N x M matrix, the matrix size is verified based on the number of extracted coordinate points, and the row and column indexes are reconstructed based on the relative positions of each point in the image (by calculating the convex hull of all points or performing cluster analysis). Subsequently, points in the first coordinate list are assigned one-to-one with points in the second coordinate list according to the same row and column order (e.g., starting from the top left corner (1,1)).

[0027] If the marker design incorporates unique shape or size differences as identifiers, shape feature parameters must be recorded simultaneously during the contour extraction stage, and matching should be performed based on these features to handle rotation or mirroring cases. The pairing algorithm must include consistency checks, such as checking whether the relative distance vectors between all paired points are approximately the same, to eliminate incorrect matches caused by image noise or marker damage. Any coordinate points that cannot find a reliable correspondence according to preset rules will be temporarily shelved or marked as invalid. All successfully paired coordinate point combinations are organized and stored as a set of matched point pairs.

[0028] Calculate the coordinate difference between each pair of coordinate points in the horizontal and vertical directions in the matching point pair set to obtain the original offset set; Each offset in the original offset set is compared with a preset tolerance threshold, and the offsets exceeding the tolerance threshold are removed to obtain the effective offset set; Specifically, the original offset set and a preset tolerance threshold are obtained, which includes a horizontal tolerance Tx and a vertical tolerance Ty. Each offset vector (ΔXi, ΔYi) in the original offset set is iterated over. For each offset vector, its absolute value of the horizontal component |ΔXi| is checked to see if it is not greater than Tx, and its absolute value of the vertical component |ΔYi| is checked to see if it is not greater than Ty.

[0029] An offset vector is considered valid data only if the absolute values ​​of both components are not greater than their corresponding tolerance thresholds. If the absolute value of either component exceeds its corresponding threshold, the offset is considered an outlier, stemming from local defects in the corresponding marker or random errors during image processing, and is therefore removed from the set to be processed. After traversing and filtering the entire original offset set, all offset vectors that pass the threshold check are retained and reorganized into a new data set, namely the valid offset set.

[0030] Calculate the horizontal average of all horizontal offsets and the vertical average of all vertical offsets in the effective offset set, and integrate the horizontal average and the vertical average to obtain the target offset.

[0031] The method provided in this embodiment pairs coordinate points according to a preset marking layout rule, ensuring that each microlens mark can be correctly associated with its theoretically corresponding pixel array mark. This guarantees the physical consistency and accuracy of the calculated offsets and prevents error propagation caused by incorrect matching. By calculating the coordinate differences of all matching point pairs and forming an original offset set, the specific deviation information generated at various local positions during the assembly process can be comprehensively obtained, providing detailed data support for the overall alignment status evaluation. Calculating the average value of the filtered effective offsets in the horizontal and vertical directions effectively filters out random errors and extracts the systematic components characterizing the overall translational deviation of the microlens array.

[0032] In one embodiment, the step of performing position compensation on the target offset when it is detected that the target offset does not meet the preset alignment accuracy threshold, until the target offset meets the preset alignment accuracy threshold, includes: The target offset is compared with the preset alignment accuracy threshold to determine whether the absolute value of the target offset is greater than the preset alignment accuracy threshold. When the absolute value of the target offset is greater than the preset alignment accuracy threshold, the target offset is subjected to direction compensation calculation to obtain position compensation information; Specifically, when the judgment result is greater than the preset alignment accuracy threshold, compensation needs to be performed, and direction compensation calculation is initiated, using the target offset that does not meet the threshold requirement as input. The offset direction is identified: the sign (positive or negative) of the horizontal offset component indicates whether the microlens array is offset to the left or right relative to the desired position in the horizontal direction; the sign of the vertical offset component indicates whether it is offset upwards or downwards. The core principle of position compensation is reverse compensation, that is, it requires driving the microlens array to move in the opposite direction to the current offset. Therefore, the direction of motion contained in the calculated position compensation information is exactly opposite to the actual offset direction indicated by the target offset. Next, the amount of displacement to be compensated is calculated. Theoretically, the compensation displacement should be equal to the absolute value of the target offset, but it needs to be converted to a physical displacement command value in micrometers or nanometers based on the calibration conversion coefficient from the image coordinate system to the physical displacement platform (e.g., the actual number of micrometers corresponding to each pixel). The calculation process also needs to consider the minimum step resolution of the displacement platform and perform necessary quantization on the theoretical displacement to generate drive parameters that the platform controller can directly parse and execute, such as a specific number of pulses or analog voltage values. The final generated position compensation information includes compensation direction commands and compensation displacement commands in at least two dimensions: horizontal and vertical.

[0033] The microlens array is driven to move in the opposite direction of the target offset based on the position compensation information; After the microlens array moves, the second marker image is reacquired, and the offset between the second marker image and the first marker image is calculated to update the target offset until the absolute value of the updated target offset is less than the preset alignment accuracy threshold.

[0034] The method provided in this embodiment objectively assesses whether the current alignment status meets the standard by comparing the calculated target offset with a preset accuracy threshold. This provides a clear and automated decision-making basis for whether to initiate compensation, eliminating the uncertainty of human experience-based judgment. When compensation is determined to be necessary, the directional compensation calculation based on the target offset can accurately generate the required displacement command. This command is directly related to the magnitude and direction of the deviation to be compensated, ensuring the purposefulness and accuracy of the compensation action. Based on the generated position compensation information, the microlens array is driven to move in the opposite direction of the deviation, realizing active physical correction of the identified overall positional deviation.

[0035] In one embodiment, when the target offset is detected to meet the preset alignment accuracy threshold, activating the photoelectric detection array around the pixel array and calculating the relative positional drift between the microlens array and the pixel array based on the response signal collected by the photoelectric detection array includes: The photoelectric detection array is activated to enter the working state and receives the raw signals output by each detection unit in the photoelectric detection array; The original signal is filtered according to a preset background noise threshold to obtain valid response signals, and the physical coordinates of each valid response signal are obtained. The signal deviation value is obtained by calculating the unit difference between each effective response signal and the preset reference signal. Specifically, a set of valid response signals is received, containing the real-time signal amplitude value and physical coordinates of each valid unit. A preset reference signal dataset, obtained during the calibration phase and strictly corresponding to the current operating conditions (such as light source intensity, spectrum, and integration time), is retrieved. This reference dataset also stores the reference signal amplitude of each unit under ideal alignment, using the detection unit index as the key.

[0036] The algorithm iterates through each item in the set of valid response signals. For each item, based on its detection unit index, it searches for and reads the corresponding reference signal amplitude value in the preset reference signal dataset. A scalar subtraction operation is performed: the currently acquired real-time signal amplitude value is subtracted from the found reference signal amplitude value. The result of this subtraction is the "signal deviation value" of the detection unit at the current moment. This deviation value directly reflects the change in the light flux illuminating the specific detection unit compared to the ideal alignment state due to the possible slight relative displacement between the microlens array and the pixel array. After completing the iterative calculation of all valid response signals, each valid signal data item is expanded to include a newly calculated signal deviation value in addition to the physical coordinates and real-time amplitude value.

[0037] Based on the signal deviation value and the physical coordinates, the spatial centroid is calculated to obtain the signal centroid coordinates; The relative position drift is obtained by calculating the drift position of the signal centroid coordinates and aligning them with the preset reference coordinates.

[0038] Specifically, the signal centroid coordinates are received, and the pre-calibrated and stored reference alignment coordinates are read from the parameter memory. The drift position is calculated by simple vector subtraction, calculating the horizontal drift component: subtracting the reference alignment coordinate from the current signal centroid abscissa.

[0039] Calculate the vertical drift component: Subtract the reference alignment coordinate from the current signal centroid's ordinate. These two differences together form a two-dimensional vector. The physical meaning of this vector is very clear: it directly indicates how the current microlens array has translated relative to its ideal alignment position in a plane parallel to the chip surface. For example, a positive horizontal drift indicates that the signal centroid has moved to the right relative to the reference, which usually means that the microlens array as a whole has shifted slightly to the left, resulting in a rightward shift of the illumination distribution centroid. The magnitude of the two-dimensional vector reflects the overall magnitude of the drift, while its direction (determined by the sign and ratio of the two components) reflects the specific orientation of the drift. This two-dimensional vector is the relative position drift.

[0040] In one embodiment, the step of filtering the original signal according to a preset background noise threshold to obtain valid response signals and acquiring the physical coordinates of each valid response signal includes: The intensity of each of the original signals is detected to obtain the signal intensity value; Each of the signal strength values ​​is compared with the preset background noise threshold; The preset background noise threshold is a fixed numerical threshold calibrated and set in advance through experiments. This threshold represents the statistical upper limit (e.g., the mean plus several times the standard deviation) of the inherent background noise level of the photoelectric detection array and its readout circuit under conditions of no effective light stimulation (or only ambient stray light). Any response signal mainly derived from real target illumination is expected to have an intensity value significantly higher than this threshold.

[0041] If the signal strength value is greater than the preset background noise threshold, then the corresponding original signal is marked as the valid response signal; The physical coordinates are obtained by performing coordinate matching based on the effective response signal and the preset unit coordinate relationship.

[0042] Specifically, the process iterates through each element in the valid response signals. For the currently processed element, its detection unit index is extracted. This index is used as the lookup key to search within the preset unit coordinate relationship, returning a physical coordinate data pair uniquely corresponding to that unit based on the index value. This retrieved physical coordinate data pair is added as a new attribute field to the currently valid response signal data element, or it can be packaged together with the signal strength value, original signal data, etc., into a more complete data structure. This traversal and matching process is performed on all valid signal elements in the set one by one, ensuring that each signal marked as valid is successfully associated with its physical coordinates. Finally, the output is a valid response signal enhanced with spatial location information.

[0043] The method provided in this embodiment, by performing intensity detection on each original signal, can uniformly quantify the photoresponse of each detection unit, providing a reliable and consistent quantitative basis for subsequent signal validity judgment based on a unified standard, and avoiding analysis difficulties caused by differences in signal form. By comparing each signal intensity value with a preset background noise threshold, signals excited by effective illumination can be strictly distinguished from inherent circuit noise and environmental interference, thereby ensuring that only signals with a qualified signal-to-noise ratio are used for subsequent analysis, significantly improving the purity and reliability of drift detection data. When the signal intensity value is greater than the threshold, it is marked as a valid response signal. This process realizes automated and rapid screening and classification of massive detection unit data, efficiently focusing on key data and reducing the computational resource consumption and time delay caused by invalid data processing.

[0044] In one embodiment, the step of calculating the spatial centroid based on the signal deviation value and the physical coordinates to obtain the signal centroid coordinates includes: The physical coordinates and signal deviation values ​​associated with each valid response signal are read sequentially. Multiply the horizontal coordinate of each physical coordinate by the corresponding signal deviation value, and sum the resulting horizontal weighted values ​​to obtain a horizontal weighted sum; Multiply the ordinate of each physical coordinate by the corresponding signal deviation value, and sum the resulting longitudinal weighted values ​​to obtain a longitudinal weighted sum; The total deviation is obtained by summing all the signal deviation values. Divide the horizontal weighted sum by the total deviation sum to obtain the horizontal coordinate of the signal centroid; Divide the weighted sum of the longitudinal axes by the total sum of deviations to obtain the longitudinal coordinate of the signal centroid; By integrating the horizontal coordinate of the signal centroid and the vertical coordinate of the signal centroid, the signal centroid coordinates are obtained.

[0045] The method provided in this embodiment ensures that all spatial and optical attribute data used for calculation are accurately and orderly loaded by sequentially reading the physical coordinates and deviation values ​​of each valid response signal. This provides a reliable and complete data source for subsequent accurate weighted calculations, avoiding omissions or confusion in data processing. By multiplying each coordinate component with its corresponding signal deviation value and summing them separately to obtain the horizontal and vertical weighted sums, and simultaneously summing all deviation values ​​to obtain the total deviation sum, this process weights and fuses discrete light intensity change signals according to their spatial location. This effectively extracts the overall trend and intensity sum of light intensity distribution changes in space, laying a mathematical foundation for calculating the distribution center.

[0046] In one embodiment, the step of dynamically aligning and maintaining the microlens array and the pixel array based on the relative position drift amount, and generating an alignment accuracy report, includes: The relative position drift is analyzed to obtain the drift direction and drift amplitude, and the drift amplitude is compared with a preset drift threshold. Specifically, the signs of the two components in the relative position drift are analyzed: the sign of the horizontal component indicates whether the horizontal offset is to the left or right; the sign of the vertical component indicates whether the vertical offset is upward or downward. The combination of these two signs (e.g., positive horizontal + positive vertical) constitutes a preliminary quadrant-based direction description. In some implementations, the angle of the vector relative to the horizontal reference axis is further calculated using the arctangent function to obtain a more precise direction angle. The values ​​of the horizontal drift component and the vertical drift component are squared, the two squared values ​​are added, and then the square root of the sum is taken. The non-negative scalar result obtained after square rooting is the drift amplitude. After extracting the direction and amplitude, a threshold comparison is performed. The calculated drift amplitude value is directly compared with the preset drift threshold value read from the parameter storage unit. It is determined whether the drift amplitude is greater than the preset threshold. This comparison operation produces a judgment result, which is a decision signal: if it is true, it indicates that the currently detected drift has exceeded the allowable stability range and compensation action needs to be triggered; if it is false, it indicates that the current alignment state is stable and recording or monitoring can be entered.

[0047] When the drift amplitude exceeds the drift threshold, a preset compensation mapping relationship is queried based on the drift direction to obtain the position compensation direction. The position compensation direction is then used to construct a compensation control signal by compensating for the drift amplitude. Specifically, when the result indicates that the drift amplitude exceeds the drift threshold, compensation is triggered. First, direction information is processed, using the drift direction as the query key. Then, a preset compensation mapping relationship is accessed. This preset mapping relationship defines a one-to-one correspondence between the drift direction and the required compensation motion direction. The typical mapping rule is reverse direction correspondence; for example, if the detected drift direction is upper right, the position compensation direction obtained through the mapping query should be lower left. Through this query, a position compensation direction instruction description matching the current drift situation is obtained.

[0048] The compensation construction phase begins, which combines directional commands with quantization information. The drift amplitude is read, reflecting the total offset to be compensated. Based on the overall spatial calibration parameters (i.e., the ratio between the movement of the photoelectric signal's centroid coordinates and the physical displacement), the drift amplitude is converted into an equivalent theoretical compensation displacement. The position compensation direction is then integrated with this theoretical compensation displacement. Specific control data is generated according to the command protocol of the underlying drive mechanism. For example, for a multi-axis motion platform, commands can be constructed as data pairs containing the X-axis target displacement increment and the Y-axis target displacement increment; for vector composite motion, commands can include the target direction angle and movement distance. The final data packet containing a clear motion direction and displacement amount is the compensation control signal.

[0049] The microlens array and the pixel array are aligned and compensated according to the compensation control signal, and the new relative position drift is re-acquired for detection. When the drift amplitude is less than the drift threshold, the current alignment state is recorded, and an alignment accuracy report is generated based on the previous compensation records.

[0050] The method provided in this embodiment analyzes the direction and magnitude of relative position drift and compares it with a preset threshold to determine in real time whether the alignment state is unstable. This provides an immediate decision-making basis for whether dynamic intervention is needed, thereby achieving proactive monitoring of minute offsets. When the drift exceeds the limit, by querying a preset mapping relationship and constructing a compensation control signal, the actuator can be precisely guided to perform reverse compensation with the correct direction and appropriate magnitude, effectively offsetting the offset that has occurred and maintaining stable alignment accuracy. Based on the compensation signal, the array is driven to realign and the drift is immediately re-detected, forming a closed-loop control cycle, so that any residual error or new drift can be continuously corrected. When the drift is below the threshold, the status is recorded and a report is generated based on previous compensations, which can completely record the alignment performance data and compensation history during the process.

[0051] Reference Figure 2 As shown, the present invention also provides a microlens and pixel array alignment device, applied to the microlens and pixel array alignment method described in any one of the above claims, comprising: The acquisition module is used to acquire a first marker image of the pixel array and a second marker image of the microlens array in the CMOS sensor chip, and to perform offset calculation on the first marker image and the second marker image to obtain the target offset. The analysis module is used to perform position compensation on the target offset when it is detected that the target offset does not meet the preset alignment accuracy threshold, until the target offset meets the preset alignment accuracy threshold. The association module is used to activate the photoelectric detection array around the pixel array when the target offset is detected to meet the preset alignment accuracy threshold, and calculate the relative position drift between the microlens array and the pixel array based on the response signal collected by the photoelectric detection array. The processing module is used to dynamically align and maintain the microlens array and the pixel array based on the relative position drift amount, and generate an alignment accuracy report.

[0052] The microlens and pixel array alignment device provided by this invention can accurately identify misalignment errors between the pixel array and microlens array in a CMOS sensor chip by acquiring marked images and calculating offsets. This provides a reliable data foundation for high-precision alignment and effectively improves the accuracy of initial positioning. By iteratively compensating for positional errors when the offset does not meet the accuracy threshold until the preset requirements are met, deviations caused by assembly and environmental factors can be gradually corrected, enhancing the stability and repeatability of the alignment process and reducing the need for manual intervention. By activating a photoelectric detection array to monitor drift in real time and dynamically maintaining alignment during the bonding stage, drift caused by material deformation and curing shrinkage can be suppressed, ensuring the reliability of alignment accuracy during long-term operation, thereby improving imaging uniformity and product performance consistency.

[0053] Reference Figure 3 As shown, the present invention also provides a microlens and pixel array alignment system, comprising: Memory, used to store programs; A processor is configured to execute the program to implement the various steps of the microlens and pixel array alignment method described in any of the preceding claims.

[0054] The present invention also provides a storage medium storing computer instructions for causing a computer to perform the method according to any one of the preceding claims.

[0055] In this embodiment, the processor and memory can be connected via a bus or other means. The memory may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive. The processor may be a general-purpose processor, such as a central processing unit, digital signal processor, application-specific integrated circuit, or one or more integrated circuits configured to implement embodiments of the present invention.

[0056] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the system and each module described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0057] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for aligning a microlens with a pixel array, characterized in that, include: The first marker image of the pixel array and the second marker image of the microlens array in the CMOS sensor chip are acquired, and the offset of the first marker image and the second marker image is calculated to obtain the target offset. When the target offset is detected to be not satisfied with the preset alignment accuracy threshold, position compensation is performed on the target offset until the target offset satisfies the preset alignment accuracy threshold. When the target offset is detected to meet the preset alignment accuracy threshold, the photoelectric detection array around the pixel array is activated, and the relative position drift between the microlens array and the pixel array is calculated based on the response signal collected by the photoelectric detection array. The microlens array and the pixel array are dynamically aligned and maintained based on the relative position drift, and an alignment accuracy report is generated.

2. The microlens and pixel array alignment method according to claim 1, characterized in that, The acquisition of a first marked image of the pixel array and a second marked image of the microlens array in the CMOS sensor chip, and the calculation of the offset between the first marked image and the second marked image to obtain the target offset, includes: The center coordinates of the markers are extracted from the first marker image and the second marker image respectively, resulting in a first coordinate list and a second coordinate list; According to the preset marker layout, the first coordinate list and the second coordinate list are paired to form a set of matching point pairs; Calculate the coordinate difference between each pair of coordinate points in the horizontal and vertical directions in the matching point pair set to obtain the original offset set; Each offset in the original offset set is compared with a preset tolerance threshold, and the offsets exceeding the tolerance threshold are removed to obtain the effective offset set; Calculate the horizontal average of all horizontal offsets and the vertical average of all vertical offsets in the effective offset set, and integrate the horizontal average and the vertical average to obtain the target offset.

3. The microlens and pixel array alignment method according to claim 1, characterized in that, The step of performing position compensation on the target offset when it is detected that the target offset does not meet the preset alignment accuracy threshold, until the target offset meets the preset alignment accuracy threshold, includes: The target offset is compared with the preset alignment accuracy threshold to determine whether the absolute value of the target offset is greater than the preset alignment accuracy threshold. When the absolute value of the target offset is greater than the preset alignment accuracy threshold, the target offset is subjected to direction compensation calculation to obtain position compensation information; The microlens array is driven to move in the opposite direction of the target offset based on the position compensation information; After the microlens array moves, the second marker image is reacquired, and the offset between the second marker image and the first marker image is calculated to update the target offset until the absolute value of the updated target offset is less than the preset alignment accuracy threshold.

4. The microlens and pixel array alignment method according to claim 1, characterized in that, When the target offset is detected to meet the preset alignment accuracy threshold, the photoelectric detection array around the pixel array is activated, and the relative positional drift between the microlens array and the pixel array is calculated based on the response signal collected by the photoelectric detection array, including: The photoelectric detection array is activated to enter the working state and receives the raw signals output by each detection unit in the photoelectric detection array; The original signal is filtered according to a preset background noise threshold to obtain valid response signals, and the physical coordinates of each valid response signal are obtained. The signal deviation value is obtained by calculating the unit difference between each effective response signal and the preset reference signal. Based on the signal deviation value and the physical coordinates, the spatial centroid is calculated to obtain the signal centroid coordinates; The relative position drift is obtained by calculating the drift position of the signal centroid coordinates and aligning them with the preset reference coordinates.

5. The microlens and pixel array alignment method according to claim 4, characterized in that, The step of filtering the original signal according to a preset background noise threshold to obtain valid response signals and acquiring the physical coordinates of each valid response signal includes: The intensity of each of the original signals is detected to obtain the signal intensity value; Each signal strength value is compared with the preset background noise threshold; If the signal strength value is greater than the preset background noise threshold, then the corresponding original signal is marked as the valid response signal; The physical coordinates are obtained by performing coordinate matching based on the effective response signal and the preset unit coordinate relationship.

6. The microlens and pixel array alignment method according to claim 4, characterized in that, The step of calculating the spatial centroid based on the signal deviation value and the physical coordinates to obtain the signal centroid coordinates includes: The physical coordinates and signal deviation values ​​associated with each valid response signal are read sequentially. Multiply the horizontal coordinate of each physical coordinate by the corresponding signal deviation value, and sum the resulting horizontal weighted values ​​to obtain a horizontal weighted sum; Multiply the ordinate of each physical coordinate by the corresponding signal deviation value, and sum the resulting longitudinal weighted values ​​to obtain a longitudinal weighted sum; The total deviation is obtained by summing all the signal deviation values. Divide the horizontal weighted sum by the total deviation sum to obtain the horizontal coordinate of the signal centroid. Divide the weighted sum of the longitudinal axes by the total sum of deviations to obtain the longitudinal coordinate of the signal centroid; By integrating the horizontal coordinate of the signal centroid and the vertical coordinate of the signal centroid, the signal centroid coordinates are obtained.

7. The microlens and pixel array alignment method according to claim 1, characterized in that, The step of dynamically aligning and maintaining the microlens array and the pixel array based on the relative position drift, and generating an alignment accuracy report, includes: The relative position drift is analyzed to obtain the drift direction and drift amplitude, and the drift amplitude is compared with a preset drift threshold. When the drift amplitude exceeds the drift threshold, a preset compensation mapping relationship is queried based on the drift direction to obtain the position compensation direction. The position compensation direction is then used to construct a compensation control signal by compensating for the drift amplitude. The microlens array and the pixel array are aligned and compensated according to the compensation control signal, and the new relative position drift is re-acquired for detection. When the drift amplitude is less than the drift threshold, the current alignment state is recorded, and an alignment accuracy report is generated based on the previous compensation records.

8. A microlens and pixel array alignment device, characterized in that, The microlens and pixel array alignment method applied to any one of claims 1-7 includes: The acquisition module is used to acquire a first marker image of the pixel array and a second marker image of the microlens array in the CMOS sensor chip, and to perform offset calculation on the first marker image and the second marker image to obtain the target offset. The analysis module is used to perform position compensation on the target offset when it is detected that the target offset does not meet the preset alignment accuracy threshold, until the target offset meets the preset alignment accuracy threshold. The association module is used to activate the photoelectric detection array around the pixel array when the target offset is detected to meet the preset alignment accuracy threshold, and calculate the relative position drift between the microlens array and the pixel array based on the response signal collected by the photoelectric detection array. The processing module is used to dynamically align and maintain the microlens array and the pixel array based on the relative position drift amount, and generate an alignment accuracy report.

9. A microlens and pixel array alignment system, characterized in that, include: Memory, used to store programs; A processor for executing the program to implement the steps of the microlens and pixel array alignment method as described in any one of claims 1-7.

10. A storage medium, characterized in that, The computer contains computer instructions for causing the computer to perform the method according to any one of claims 1 to 7.

Images