Adaptive Optimization Method for 3D DIC Parameters Based on Image Quality Assessment
By using an adaptive optimization method based on image quality assessment, the 3D DIC parameters are automatically adjusted, solving the problem of parameter setting relying on human experience, improving the reliability and consistency of measurement results, and achieving efficient image quality assessment and optimization.
Patent Information
- Application Number
- CN202511964786.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-24
- Publication Date
- 2026-05-26
AI Technical Summary
The parameter settings of existing 3D DIC technology rely on human experience, which is inefficient and makes it difficult to guarantee the consistency and accuracy of measurement results. Furthermore, existing image quality indicators have limitations in evaluation, resulting in poor image quality and affecting the reliability of measurement results.
An adaptive optimization method based on image quality assessment is adopted. By using multi-dimensional image quality indicators and the CRITIC method for quantitative assessment, combined with Taguchi experiments and Bayesian optimization algorithms, camera and system parameters are automatically adjusted to achieve accurate assessment and optimization of image quality.
It achieves automated optimization of 3D DIC parameters, improves the reliability and consistency of measurement results, reduces manual debugging costs and time, and enhances the system's environmental robustness and operational stability.
Smart Images
Figure CN122093655A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of machine vision 3D measurement and mechanical testing technology, and specifically relates to an adaptive optimization method and system for acquisition parameters in digital image correlation methods. Background Technology
[0002] Digital Image Correlation (DIC), particularly 3D DIC, is a non-contact optical deformation measurement technique. It reconstructs the three-dimensional shape and displacement field of an object by tracking changes in the speckle pattern on its surface before and after deformation. This technique has been widely applied in experimental research in fields such as materials mechanics, structural engineering, and biomedicine. However, the measurement accuracy of 3D DIC technology is highly dependent on the quality of the acquired image sequence. High-quality images should possess the following characteristics: clear speckle patterns, high contrast, no overexposure or underexposure, low noise, and no artifacts such as specular reflection. To obtain such images, users must precisely set a series of complex acquisition parameters, primarily including camera acquisition parameters such as aperture value, exposure time, and frame rate; and system parameters such as camera working distance and light source brightness.
[0003] Currently, parameter settings for 3D DIC technology heavily rely on operator experience, requiring a tedious process of trial and error, repeated adjustments, and subjective visual judgment of image quality. This method is not only inefficient and time-consuming, but also, because the human eye cannot perform quantitative evaluation, it is difficult to find the globally optimal parameter combination. This leads to significant differences in settings between different operators or experiments, severely affecting the consistency and scientific validity of measurement results. However, even with automated methods, existing common image quality metrics (such as signal-to-noise ratio and information entropy) still have inherent limitations in evaluating 3D DIC images. For example, a set of images may have a high average gradient, but its speckle pattern may have local clusters or gaps, easily causing "speckles" or "jumping points" in deformation calculations. Furthermore, when facing highly reflective metal surfaces, simple contrast optimization is insufficient to effectively suppress saturation areas caused by specular reflection, resulting in the loss of speckle information and compromising the integrity of displacement field calculations. Summary of the Invention
[0004] To address the problems of low efficiency, difficulty in guaranteeing accuracy, and poor repeatability in parameter setting in existing technologies, the present invention aims to provide a 3D DIC parameter adaptive optimization method and system based on image quality assessment. This method and system can accurately assess the traceability and computational robustness of 3D DIC images and integrate automated hardware control functions to achieve true automation and intelligence in 3D DIC parameter optimization.
[0005] The objective of this invention is achieved through the following technical solution:
[0006] The 3D DIC parameter adaptive optimization method based on image quality assessment disclosed in this invention includes the following steps:
[0007] Step 1: Parameter initialization.
[0008] The parameters include camera acquisition parameters and system parameters.
[0009] Camera parameters include aperture value, exposure time, and frame rate.
[0010] System parameters include camera working distance and light source brightness.
[0011] The above parameter combinations are assigned to multiple sets of empirical values.
[0012] Step 2: Image acquisition step. Based on the current combination of multiple parameters, the acquisition device is controlled to acquire multiple images of the object under test.
[0013] Step 3: Multi-camera synchronization detection.
[0014] 3.1 Select any three sets of images from the images acquired in step two, obtain the system timestamps of the images, and calculate the synchronization index Δt_sync of the multiple cameras:
[0015] Δt_sync = max(|t_i - t_j|)
[0016] t_i and t_j are the image timestamps of the i-th and j-th cameras recording the same physical event.
[0017] 3.2 Determine whether the multi-camera system is synchronized.
[0018] For static measurement tasks, whether triggered by software or hardware, the synchronization index Δt_sync needs to reach the millisecond level.
[0019] For dynamic measurement tasks, hardware triggering is required. The synchronization index Δt_sync is determined based on the exposure time and the characteristic time of the motion of the object being measured, and should be at least less than 1 / 10 of the exposure time.
[0020] Step 4: If the camera synchronization meets the requirements, proceed to Step 5. If the multi-camera synchronization does not meet the requirements, for software triggering, optimize the synchronization by reducing the image ROI or switch to hardware triggering. For hardware triggering, check if the camera cable connection lengths are consistent, if the interfaces are loose, and use a multi-functional synchronizer designed for synchronized imaging to optimize hardware triggering synchronization. Repeat steps 2 and 3 until the synchronization meets the requirements.
[0021] Step 5: Perform quantitative analysis on the image sequences acquired under multi-camera synchronization, and evaluate the image quality through image quality assessment indicators.
[0022] The image quality assessment metrics include image signal-to-noise ratio, image grayscale histogram entropy, image contrast, speckle distribution uniformity, average speckle size, and mean image grayscale gradient.
[0023] The speckle distribution uniformity index is obtained through the following method:
[0024] For a speckle image I(x,y), the pixel coordinates of all speckles are extracted by image processing threshold segmentation, denoted as the point set P = (p1,p2,…p…). N ), where p is the coordinate of the speckle and N is the total number of speckles. Calculate the Voronoi segmentation of the point set P across the entire image plane. Each Voronoi unit V i Defined as Where d(·) is the Euclidean distance. Each Voronoi element V i The area is A i (Unit: pixels) 2 ). Calculate the area of all cells {A1, A2, ... A N variance The speckle distribution uniformity index is defined as follows:
[0025] The image evaluation indicators are weighted using the CRITIC method, and the corresponding image quality evaluation results are calculated based on the obtained indicator weights to complete the evaluation.
[0026] Step Six: Parameter Optimization Judgment Step. Determine whether the evaluation indicators corresponding to multiple parameter combinations in Step Five meet the image quality evaluation requirements. If they meet the requirements, continue to acquire 3D DIC images with that parameter combination. If none of them meet the requirements, proceed to Step Seven for parameter optimization.
[0027] Step 7, the parameter optimization step, is based on the current multiple sets of image parameter combinations and quality assessment indicators. A new set of parameter combinations is generated through the optimization algorithm. The parameter combination is assigned to Step 1, and Steps 2, 5, and 6 are repeated. If the image quality assessment result is significantly improved but still does not meet the requirements, it is added to the parameter combination set. Steps 2, 5, 6, and 7 are repeated to continue the optimization until the image quality meets the standard. The current optimal parameter combination is output, and 3D DIC image acquisition continues.
[0028] Furthermore, the optimization algorithm described in step seven includes the following stages:
[0029] Global screening stage: Using the Taguchi method, based on a pre-designed orthogonal array, a small number of experiments were conducted within the initial value range of all parameters. By analyzing the main effects of each parameter, the key parameters that significantly affect image quality and their preliminary optimal value ranges were determined.
[0030] Local optimization stage: Using the preliminary optimal value interval as a new parameter search space, Bayesian optimization method is adopted, with the overall image quality score as the objective function, and iterative guided experiments are carried out until convergence to obtain the globally optimal parameter combination.
[0031] This invention discloses a 3D DIC parameter adaptive optimization system based on image quality assessment, used to implement the aforementioned 3D DIC parameter adaptive optimization method based on image quality assessment. The 3D DIC parameter adaptive optimization system based on image quality assessment disclosed in this invention includes a parameter initialization module, a synchronization detection and control module, an image acquisition control module, an image quality assessment module, a parameter optimization module, and a parameter output module.
[0032] The parameter initialization module is used to set the initial parameters for the acquisition equipment of the 3D DIC system.
[0033] The synchronization detection and control module is used to perform multi-camera synchronization detection and ensure that synchronization meets the standards.
[0034] The image acquisition control module is used to control the acquisition device to acquire images of the object under test based on the current parameter combination.
[0035] The image quality assessment module is used to perform quantitative analysis on the acquired images and calculate image quality assessment indicators.
[0036] The parameter optimization module is used to generate a new set of parameters based on the current parameter combination and the image quality evaluation index set when the image quality does not meet the requirements, and to trigger the image acquisition control module to start working again.
[0037] The parameter output module is used to output the current optimal parameter combination when the image quality meets the requirements.
[0038] Furthermore, it also includes a hardware unit that is communicatively connected to the image acquisition and control module. The hardware unit includes at least two cameras and lenses, a light source, an integrated displacement stage driven by a motor, and a multispectral ambient light sensor.
[0039] At least two cameras and lenses are provided, and the aperture, exposure time, and frame rate of the camera acquisition system can be adjusted by the image acquisition control module.
[0040] The brightness of the light source can be adjusted by the image acquisition control module.
[0041] An integrated displacement stage driven by a motor is used to support the camera, lens and light source. The image acquisition control module can automatically drive the platform to adjust the working distance of the camera according to the size of the object being measured.
[0042] A multispectral ambient light sensor is used to monitor ambient light intensity and color temperature in real time and feed the data back to the parameter initialization module as a priori input for parameter initialization.
[0043] Furthermore, a multispectral ambient light sensor is communicatively connected to the control unit for monitoring ambient light information.
[0044] The control unit is configured to:
[0045] (a) Before starting the parameter optimization process as described in claim 1, a preset initial reference value is provided for adjusting the light source brightness based on the information monitored by the ambient light sensor.
[0046] (b) When the system continuously collects data based on the optimal parameter set, it continuously compares the deviation between the ambient light information and the initial ambient light. If the deviation continues to exceed the preset tolerance, a calibration prompt or interruption signal is generated.
[0047] The present invention discloses a 3D DIC parameter adaptive optimization device, which includes a processor and a memory.
[0048] The memory stores computer programs.
[0049] When the processor executes the computer program, it performs the following steps:
[0050] Parameter initialization steps: Set the initial parameter combination for the acquisition device of the 3D DIC system.
[0051] Synchronization detection and control steps: Perform multi-camera synchronization detection and ensure that synchronization meets the standards.
[0052] Image acquisition control steps: Based on the current parameter combination, control the acquisition device to acquire an image of the object being measured.
[0053] Image quality assessment steps: Perform quantitative analysis on the acquired images and calculate image quality assessment indicators.
[0054] Parameter optimization steps: When the image quality does not meet the requirements, based on the current parameter combination and the image quality evaluation results, a new parameter combination is generated using the optimization algorithm described in claim 1 or 2, and the image acquisition control and evaluation are re-executed.
[0055] Parameter output steps: When the image quality meets the requirements, output the current optimal parameter combination.
[0056] Beneficial effects:
[0057] 1. This invention provides a quantifiable comprehensive scoring standard for the traceability and computational robustness of 3D DIC images for the first time by constructing a quantitative evaluation model based on multi-dimensional image quality indicators and the CRITIC objective weighting method. This solves the problem of relying on subjective experience and lacking consistent evaluation criteria in traditional parameter settings.
[0058] 2. This invention constructs an adaptive closed-loop acquisition system that integrates synchronous control, ambient light sensing, and hardware linkage. It can dynamically adjust acquisition conditions under complex working conditions, thereby effectively enhancing the environmental robustness, operational stability, and overall automation level of the 3D DIC system.
[0059] 3. This invention proposes an image quality assessment method based on multi-index fusion and objective weighting. By integrating key indicators such as image signal-to-noise ratio and speckle uniformity and using the CRITIC method to determine weights, it achieves accurate quantitative assessment of image traceability and computational robustness, providing a reliable basis for parameter optimization.
[0060] 4. This invention designs a two-stage intelligent optimization process of "screening-optimization". First, the Taguchi experimental method is used to quickly identify the effective range of key parameters. Then, Bayesian optimization is used to perform efficient global optimization within this range. Finally, through the automated closed loop of "evaluation-optimization", the optimal parameter combination is directly output. While ensuring the reliability of the results, the cost and time of manual debugging are greatly reduced. Attached Figure Description
[0061] Figure 1 This is a flowchart of a 3D DIC parameter adaptive optimization method based on image quality assessment.
[0062] Figure 2 This is a schematic diagram of a 3D DIC parameter adaptive optimization system based on image quality assessment. Detailed Implementation
[0063] To better illustrate the purpose and advantages of the present invention, the invention will be further described below in conjunction with the accompanying drawings and examples.
[0064] Example 1:
[0065] This embodiment takes the measurement of the full-field strain field of an aerospace aluminum alloy sheet during the tensile process as an example.
[0066] 1. System Hardware Configuration
[0067] like Figure 2As shown in this embodiment, the 3D DIC parameter adaptive optimization system based on image quality assessment includes a parameter initialization module, a synchronization detection and control module, an image acquisition control module, an image quality assessment module, a parameter optimization module, and a parameter output module.
[0068] Image acquisition unit: Two industrial cameras with a resolution of 4096×3000 pixels, equipped with 16mm fixed-focus lenses. The cameras are connected to the control computer via gigabit Ethernet ports.
[0069] Illumination unit: Two high-brightness white LED strip light sources illuminate the surface of the object being measured symmetrically at an angle of approximately 30 degrees. The brightness can be continuously adjusted via a PWM signal.
[0070] Synchronization and control unit: a multi-functional synchronizer for sending hardware external trigger signals to the two cameras; an industrial computer as the main controller.
[0071] Auxiliary sensor (optional): A multispectral ambient light sensor installed near the measurement area to monitor ambient illuminance (in Lux) and color temperature.
[0072] Test object: An aluminum alloy sheet with a black and white matte speckle pattern sprayed on its surface, with an average speckle diameter of about 4-5 pixels.
[0073] like Figure 1 As shown in this embodiment, the adaptive optimization of 3D DIC parameters based on image quality assessment includes the following steps:
[0074] Step 1: Parameter Initialization and System Settings
[0075] The parameter initialization module in the main controller loads a preset combination of empirical parameters for "tensile measurement of metallic materials". Five sets of empirical values are assigned to the initial parameters, as shown in the example below:
[0076] Parameter combination number Aperture (F) Exposure time (μs) Light source brightness (%) Working distance (mm) 1 5.6 8000 30 400 2 5.6 12000 50 400 3 8.0 8000 50 400 4 8.0 12000 30 400 5 6.6 10000 40 400
[0077] If the system is configured and a multispectral ambient light sensor is enabled, its data will be deeply integrated into the parameter optimization process, providing dynamic constraints on the light source brightness. First, the system reads the current ambient light illuminance (e.g., 300 Lux) and, based on the preset optimal object surface illuminance range (e.g., 1400-1600 Lux), calculates the currently allowed active light source brightness adjustment range (e.g., 30%-50%), and sets it as the initial search range for the brightness parameters.
[0078] In subsequent Taguchi method experiments, the level setting of the brightness parameter will be selected within this range; in the Bayesian optimization iteration, this range will serve as the dynamic search boundary for the brightness parameter. This mechanism ensures that the system can automatically meet the basic imaging illumination requirements and find the globally optimal parameter combination when optimizing in any environment.
[0079] Step 2: Image Trial Acquisition and Synchronization Detection
[0080] Control two cameras and take static photos sequentially according to the five sets of parameters in the table above, capturing 10 frames of images for each set of parameters.
[0081] Synchronization detection: The controller selects one frame of image acquired under the same hardware trigger pulse for each set of parameters and reads its hardware timestamp. For example, for parameter combination 1, the timestamp of camera A is measured to be 1000.001200s, and that of camera B is 1000.001050s. The synchronization index Δt_sync is calculated using the formula: Δt_sync = |1000.001200 - 1000.001050| = 0.00015s (150μs).
[0082] Judgment and Optimization: This embodiment involves dynamic strain measurement, requiring high synchronization accuracy. A threshold is set to 1 / 10 of the target exposure time (e.g., if the target exposure time is 10ms, then the threshold is 1ms). A measured value of 150μs is far less than the threshold, indicating that synchronization is achieved. If not, hardware checks and optimizations are performed according to the steps described in claim 1.
[0083] Step 3: Image Quality Assessment
[0084] For images that meet the synchronization standards, the quality assessment module calculates six indicators. Taking an image acquired with parameter combination 1 as an example:
[0085] Speckle distribution uniformity: Approximately 15,000 speckle spots were detected, a Voronoi diagram was generated, and the variance of the area of all cells was calculated. The variance σ_A was then calculated. 2 The smaller size indicates that the speckle pattern is evenly distributed.
[0086] Other metrics: Simultaneously calculate the image's signal-to-noise ratio (SNR = 24.5 dB), image entropy (H = 7.2), contrast ratio (C = 0.35), average speckle size (4.2 pixels), and average grayscale gradient (G = 12.5).
[0087] Overall Score: Based on the data from all 5 sets of parameter images in this experiment, the objective weights of each indicator were calculated using the CRITIC weighting method. The calculated weight vector is W = [0.25, 0.10, 0.15, 0.30, 0.10, 0.10] (corresponding to the above 6 indicators respectively). Therefore, the overall score S1 of parameter combination 1 is calculated as: normalized value of each indicator · weight vector W.
[0088] Step 4: Parameter Optimization Judgment and Execution
[0089] The system's preset comprehensive score threshold S_threshold = 0.85. Calculations showed that the highest comprehensive score for the five initial parameter sets was 0.78, which did not meet the target.
[0090] Triggering phased parameter optimization:
[0091] Phase 1 - Global Screening Using the Taguchi Method:
[0092] Select four key parameters (aperture, exposure time, light source brightness, and working distance), each with three levels, and set them according to L9 (3 4 Nine experiments were arranged using an orthogonal array.
[0093] Nine experiments were conducted, and the overall score for each experiment was calculated. The main effect plot was derived by analyzing the relationship between the signal-to-noise ratio (S / N) and the levels of each parameter. The analysis showed that exposure time and light source brightness were significant factors affecting the overall score, with optimal levels in the ranges of [10000, 15000] μs and [60%, 80%] brightness, respectively. Aperture and working distance had no significant impact within the test range.
[0094] Phase Two - Bayesian Optimization for Local Finding:
[0095] Using the aforementioned optimal interval as the new search boundary, and fixing other parameters at the intermediate level, Bayesian optimization is initiated with the overall score S as the objective function.
[0096] The optimizer converged after 20 iterations. During the iteration process, the system automatically executed a closed loop of image acquisition, evaluation, and feedback. The final recommended optimal parameters were: aperture F6.4, exposure time 12500μs, light source brightness 72%, and working distance 800mm. Under these parameters, the overall score S = 0.92 was obtained, which meets the requirements.
[0097] Step 5: Application and Verification of Optimal Parameters
[0098] The system outputs and locks the optimal parameter combination, and controls the DIC system to start formal data acquisition with these parameters to perform tensile tests on the aluminum alloy sheet.
[0099] During the one-hour continuous data acquisition process, the ambient light sensor detected occasional flickering of indoor fluorescent lights, causing the ambient illuminance to fluctuate between 280 and 320 Lux. Since active light source compensation was locked in step one, this fluctuation did not change the effective illuminance of the object surface, and the system continued to output high-quality images stably without triggering any warnings.
[0100] 3. Implementation Results
[0101] In this embodiment, the system automatically completed the entire process from parameter initialization, synchronization verification, image quality assessment to intelligent optimization within approximately 30 minutes, ultimately obtaining optimal acquisition parameters (overall score 0.92) that were superior to those set manually based on experience (overall score 0.78). Applying these parameters to actual dynamic strain measurement improved the signal-to-noise ratio of the calculated DIC data, effectively ensuring the reliability of high-precision strain measurement results.
[0102] The above detailed description further illustrates the purpose, technical solution, and beneficial effects of the invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A 3DDIC parameter adaptive optimization method based on image quality assessment, characterized in that, Includes the following steps: Step 1: Parameter initialization; The parameters include camera acquisition parameters and system parameters; Camera capture parameters include aperture value, exposure time, and frame rate; System parameters include camera working distance and light source brightness; The above parameter combinations are assigned to multiple sets of empirical values; Step 2: Image acquisition step, based on the current combination of multiple parameters, control the acquisition device to acquire multiple sets of images of the object under test; Step 3: Multi-camera synchronization detection; 3.1 Select any three sets of images from the images acquired in step two, obtain the system timestamps of the images, and calculate the synchronization index Δt_sync of the multiple cameras: Δt_sync = max(|t_i - t_j|) t_i and t_j are the image timestamps of the i-th and j-th cameras recording the same physical event; 3.2 Determine if the multi-camera system is synchronized; For static measurement tasks, whether triggered by software or hardware, the synchronization index Δt_sync needs to reach the millisecond level. For dynamic measurement tasks, hardware triggering is required. The synchronization index Δt_sync is determined based on the exposure time and the characteristic time of the motion of the object being measured, and should be at least less than 1 / 10 of the exposure time. Step 4: If the camera synchronization meets the requirements, proceed to Step 5; if the multi-camera synchronization does not meet the requirements, optimize the software trigger synchronization by reducing the image ROI or switch to hardware triggering. For hardware triggering, check whether the camera cable connection length is consistent, whether the interface is loose, and use a multi-functional synchronizer designed for synchronous imaging to optimize the hardware trigger synchronization. Repeat steps two and three until the synchronization meets the standard. Step 5: Perform quantitative analysis on the image sequences acquired under multi-camera synchronization, and evaluate the image quality through image quality assessment indicators; The image quality assessment metrics include image signal-to-noise ratio, image gray-level histogram entropy, image contrast, speckle distribution uniformity, average speckle size, and image gray-level gradient mean. The image evaluation indicators are weighted using the CRITIC method, and the corresponding image quality evaluation results are calculated based on the obtained indicator weights to complete the evaluation. Step Six: Parameter Optimization Judgment Step. Determine whether the evaluation indicators corresponding to multiple parameter combinations in Step Five meet the image quality evaluation requirements. If they meet the requirements, continue to acquire 3DDIC images with that parameter combination. If none of them meet the requirements, proceed to Step Seven: Parameter Optimization. Step 7, the parameter optimization step, is based on the current multiple sets of image parameter combinations and quality assessment indicators. A new set of parameter combinations is generated through the optimization algorithm. The parameter combination is assigned to Step 1, and Steps 2, 5, and 6 are repeated. If the image quality assessment result is significantly improved but still does not meet the requirements, it is added to the parameter combination set. Steps 2, 5, 6, and 7 are repeated to continue the optimization until the image quality meets the standard. The current optimal parameter combination is output, and 3DDIC image acquisition continues.
2. The method according to claim 1, characterized in that, In step five, The speckle distribution uniformity index was obtained using the following method: For a speckle image I(x,y), the pixel coordinates of all speckles are extracted by image processing threshold segmentation, denoted as the point set P = (p1,p2,…p…). N ), where p is the coordinate of the speckle and N is the total number of speckles; calculate the Voronoi segmentation of the point set P across the entire image plane; each Voronoi unit V i Defined as Where d(·) is the Euclidean distance; each Voronoi unit V i The area is A i ; Calculate the area of all cells {A1, A2, ... A N variance The speckle distribution uniformity index is defined as follows:
3. The 3DDIC parameter adaptive optimization method based on image quality assessment as described in claim 2, characterized in that, The optimization algorithm described in step seven includes the following stages: Global screening stage: Using the Taguchi method, based on a pre-designed orthogonal array, a small number of experiments were conducted within the initial value range of all parameters. By analyzing the main effects of each parameter, the key parameters that significantly affect image quality and their preliminary optimal value ranges were determined. Local optimization stage: Using the preliminary optimal value interval as a new parameter search space, Bayesian optimization method is adopted, with the overall image quality score as the objective function, and iterative guided experiments are carried out until convergence to obtain the globally optimal parameter combination.
4. A system for implementing the method as described in claim 1, 2, or 3, characterized in that, It includes a parameter initialization module, a synchronization detection and control module, an image acquisition control module, an image quality assessment module, a parameter optimization module, and a parameter output module; The parameter initialization module is used to set the initial parameters for the acquisition devices of the 3DDIC system. The synchronization detection and control module is used to perform multi-camera synchronization detection and ensure that synchronization meets the standards. The image acquisition control module is used to control the acquisition device to acquire images of the object under test based on the current parameter combination; The image quality assessment module is used to perform quantitative analysis on the acquired images and calculate image quality assessment indicators. The parameter optimization module is used to generate a new set of parameters based on the current parameter combination and the image quality evaluation index set when the image quality does not meet the requirements, and to trigger the image acquisition control module to work again. The parameter output module is used to output the current optimal parameter combination when the image quality meets the requirements.
5. The system as described in claim 4, characterized in that: It also includes a hardware unit that is communicatively connected to the image acquisition and control module. The hardware unit includes at least two cameras and lenses, a light source, an integrated displacement stage driven by a motor, and a multispectral ambient light sensor. At least two cameras and lenses are provided, and the aperture, exposure time, and frame rate of the camera acquisition system can be adjusted by the image acquisition control module. The brightness of the light source can be adjusted by the image acquisition control module. An integrated displacement stage driven by a motor is used to support the camera, lens and light source. The image acquisition and control module can automatically drive the platform to adjust the working distance of the camera according to the size of the object being measured. A multispectral ambient light sensor is used to monitor ambient light intensity and color temperature in real time and feed the data back to the parameter initialization module as a priori input for parameter initialization.
6. The system according to claim 5, characterized in that, The control unit is configured to: (a) Before starting the parameter optimization process as described in claim 1, a preset initial reference value is provided for adjusting the light source brightness based on the information monitored by the ambient light sensor; (b) When the system continuously collects data based on the optimal parameter set, it continuously compares the deviation between the ambient light information and the initial ambient light. If the deviation continues to exceed the preset tolerance, a calibration prompt or interruption signal is generated.
7. A 3DDIC parameter adaptive optimization device, characterized in that, Including processor and memory; The memory stores computer programs; When the processor executes the computer program, it performs the following steps: Parameter initialization steps: Set the initial parameter combination for the acquisition device of the 3DDIC system; Synchronization detection and control steps: Perform multi-camera synchronization detection and ensure that synchronization meets the standards; Image acquisition control steps: Based on the current parameter combination, control the acquisition device to acquire an image of the object under test; Image quality assessment steps: Perform quantitative analysis on the acquired images and calculate image quality assessment indicators; Parameter optimization steps: When the image quality does not meet the requirements, based on the current parameter combination and the image quality evaluation results, a new parameter combination is generated using the optimization algorithm described in claim 1 or 2, and the image acquisition control and evaluation are re-executed. Parameter output steps: When the image quality meets the requirements, output the current optimal parameter combination.