An automated detection system for in-vehicle electronic streaming rearview mirrors

Abstract

This invention discloses an automated testing system for in-vehicle electronic streaming rearview mirrors, relating to the field of streaming rearview mirror testing. The system includes a light source dark box, a multi-axis light source array, a moving light source track, a dimming module, a temperature and humidity control module, a camera, a dynamic light test environment determination module, and an output module. Based on a parallel deep random forest model, an adaptive test environment is determined, ensuring that the test conditions always focus on the sensitive range most likely to expose potential faults. This achieves a standardized dynamic light environment simulation test system, maximizing defect detection rate and fault mode recognition accuracy while ensuring test efficiency, and establishing a dynamically enhanced optical stress test system for product quality verification.

Description

Technical Field

[0001] This invention relates to the field of streaming media rearview mirror inspection, and in particular to an automated inspection system for in-vehicle electronic streaming media rearview mirrors. Background Technology

[0002] The automated testing of in-vehicle electronic streaming rearview mirrors is crucial for driving safety, directly impacting the driver's ability to assess real-time road conditions in complex environments. A key current industry weakness lies in the lack of a standardized dynamic lighting environment simulation testing system, particularly in accurately reproducing extreme lighting scenarios such as direct glare from vehicles behind at night or sudden changes in light intensity in tunnels. Existing testing processes largely rely on static lighting conditions, leading to serious defects in products under real-world conditions: for example, sudden strong light can overload the image sensor, causing glare, whitening, and loss of detail, ultimately rendering the rearview mirror incapable of basic environmental awareness. This lack of dynamic lighting environment verification means that potential optical compatibility defects are not effectively addressed, ultimately leading to misjudgments in driving decisions. Building an automated testing environment capable of simulating dynamic lighting interference has become a necessary technological foundation for improving product optical robustness and ensuring all-weather safe visibility. Summary of the Invention

[0003] To address the technical problem of the lack of standardized dynamic light environment simulation testing in existing technologies, this invention provides an automated testing system for in-vehicle electronic streaming media rearview mirrors.

[0004] This invention is achieved through the following technical solution:

[0005] An automated detection system for in-vehicle electronic streaming rearview mirrors includes:

[0006] Light source dark box, multi-axis light source array, moving light source track, dimming module, temperature and humidity control module, camera, dynamic light test environment determination module, output module;

[0007] The multi-axis light source array consists of several independently controllable LEDs, used to simulate natural light and artificial light;

[0008] The multi-axis light source array is mounted on a moving light source track and can move on the track;

[0009] The dynamic light test environment determination module includes a preliminary test environment determination module, a detection result processing module, and an adaptive test environment determination module, which are used to determine the preliminary test environment and the adaptive test environment.

[0010] The dimming module and the temperature and humidity control module adjust the dynamic light source parameters and the temperature and humidity of the test environment based on the output of the dynamic light test environment determination module.

[0011] The camera is used to capture electronic streaming rearview mirror images during testing;

[0012] The output module outputs the total test result based on the test results under the test environment determined by the dynamic light test environment determination module.

[0013] Furthermore, the dynamic light source parameters include light source optical parameters, motion parameters, and environmental parameters.

[0014] Furthermore, the optical parameters of the light source include the spectral characteristics and polarization characteristics of the light source;

[0015] The motion parameters include brightness change parameters, flashing frequency, motion trajectory, moving speed, acceleration, and angle change rate;

[0016] The environmental parameters are used for environmental coupling of simulated light, and are adjusted by regulating the temperature and humidity control module. 。

[0017] Furthermore, the dynamic light testing environment determination module specifically includes the following steps:

[0018] S1: Building the initial test environment;

[0019] The construction of the preliminary test environment includes the determination of the dynamic light source parameter library and the determination of the dynamic light source parameters in the preliminary test environment;

[0020] S2: Obtain the test results under the initial test environment and the test results of the same batch of products, and extract features;

[0021] S3: Based on the test results of the initial test environment and the test results of the same batch of products, conduct supplementary tests to establish an adaptive test environment.

[0022] Furthermore, the determination of the dynamic light source parameter library includes compensating and correcting the simulated light by combining environmental interference parameters, atmospheric transmission parameters, and surface reflection parameters, to obtain a final and complete dynamic light source parameter library output based on the actual use scenario and the specific environment.

[0023] Furthermore, the determination of dynamic light source parameters in the preliminary test environment is based on fault orientation, including acquiring historical fault detection data and classifying faults into N categories according to fault phenomena; extracting light source parameters for each category of fault, including light source optical parameters, motion parameters, and environmental parameters; and constructing a preliminary test environment based on the selection of dynamic light source parameters for testing according to the fault category.

[0024] Furthermore, the supplementary testing based on the initial test environment results and the test results of the same batch of products, to establish an adaptive test environment, includes:

[0025] S31: Establish a parallel deep random forest model;

[0026] S32: Perform model training;

[0027] S33: Input the image features of the detection results of the initial test environment and the detection results of the same batch of products, and generate the dynamic light source parameters and the number of test cases for supplementary testing based on the trained model;

[0028] S34: Based on the parameters of the dynamic light source used for supplementary testing and the number of test cases, an adaptive test environment is established to obtain the final dynamic light test environment.

[0029] Furthermore, the parallel deep random forest model comprises a three-layer forest. The first layer of the forest is composed of image features from the detection results of the initial test environment and the detection results of products from the same batch. ,in, These are the optical performance characteristics, color characteristics, geometric characteristics, texture characteristics, and temporal characteristics of the i-th sample, respectively.

[0030] Furthermore, the input to the first layer of the parallel deep random forest model is:

[0031]

[0032] Output of layer p (p=1,2,3):

[0033]

[0034] Where T is the number of trees in each layer of the forest. Let be the model function for the t-th tree in the p-th layer. For the input of the p-th layer;

[0035] The expression is as follows:

[0036]

[0037] [ ; ] represents the feature splicing operation.

[0038] Furthermore, the loss function expression of the parallel deep random forest model is as follows:

[0039]

[0040] in,

[0041] Prediction error is used to measure the difference between the model's prediction and the actual value. For testing costs; For physical constraints; For regularization terms; , , Weighting coefficients

[0042] Compared with existing technologies, the advantages of this invention are as follows: A dynamically reconfigurable light source control platform is constructed through a systematic adaptive testing environment, driven by a closed-loop decision-making mechanism based on real-time collected test feedback data. The system intelligently explores light source parameters through multi-dimensional optimization algorithms, generating highly customized light source strategies for the core pain points of different testing scenarios. Each round of testing dynamically adjusts parameter combinations based on historical performance, prioritizing the test light source parameters that are most challenging for the current test object, ensuring that test conditions always focus on the sensitive range most likely to expose potential faults. This closed-loop adaptive mechanism not only avoids the coverage blind spots of fixed light source configurations but also actively induces abnormal responses under critical conditions by continuously applying targeted optical pressure, thereby maximizing defect detection rate and fault mode recognition accuracy while ensuring testing efficiency, establishing a dynamically enhanced optical stress testing system for product quality verification. Attached Figure Description

[0043] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0044] Figure 1 This is a schematic diagram of the workflow of the dynamic light testing environment determination module according to an embodiment of this application;

[0045] Figure 2 This is a schematic diagram of the parallel deep random forest model structure according to an embodiment of this application. Detailed Implementation

[0046] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0047] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0048] It should also be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. The illustrations only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the shape, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0049] An automated detection system for in-vehicle electronic streaming rearview mirrors includes:

[0050] Light source dark box, multi-axis light source array, moving light source track, dimming module, temperature and humidity control module, camera, dynamic light test environment determination module, output module;

[0051] Specifically, the multi-axis light source array consists of several independently controllable LEDs, used to simulate natural light and artificial light;

[0052] The multi-axis light source array is mounted on a moving light source track and can move on the track;

[0053] The dynamic light test environment determination module includes a preliminary test environment determination module, a preliminary detection result processing module, and an adaptive test environment determination module. Its workflow is as follows: the preliminary test environment determination module constructs a preliminary test environment, the adaptive test environment is constructed based on historical detection results, and finally the test results under the preliminary test environment and the adaptive test environment are obtained.

[0054] The dimming module and the temperature and humidity control module determine the output parameters of the module according to the dynamic light test environment, that is, the dynamic light source parameters adjust the optical parameters, motion parameters of the light source and the environmental parameters of the test environment.

[0055] Furthermore, it also includes controlling the motion trajectory, speed, acceleration, and angular change rate of the multi-axis light source array on the moving light source track;

[0056] The camera is used to capture electronic streaming rearview mirror images during testing;

[0057] The output module outputs the total test result based on the test results under the test environment determined by the dynamic light test environment determination module.

[0058] The overall test results include the types of faults detected and the degree of deviation.

[0059] The dynamic light testing environment determination module specifically includes the following steps: Figure 1 As shown:

[0060] S1: Building the initial test environment;

[0061] The construction of the preliminary test environment includes the determination of the dynamic light source parameter library and the determination of the dynamic light source parameters in the preliminary test environment;

[0062] S11: Determination of the dynamic light source parameter library;

[0063] The determination of the dynamic light source parameter library includes selecting dynamic light source parameters based on actual conditions, including the selection of light source optical parameters, motion parameters, and environmental parameters;

[0064] The optical parameters of the light source include the spectral characteristics and polarization characteristics of the light source; the spectral characteristics include color temperature, half-width of the spectrum, color rendering index, and peak wavelength.

[0065] Motion parameters include brightness variation parameters, flicker frequency, motion trajectory, movement speed, acceleration, and angle change rate; among which, the brightness variation parameters include different types of variation modes, such as step change mode, which is used to simulate tunnel protrusion / entry, exponential decay mode, which is used to simulate sunset scenes, sinusoidal modulation mode, which is used to simulate tree-lined road strobe, and random pulse mode, which is used to simulate random vehicle headlight appearance.

[0066] Environmental parameters are used to couple the simulated light to the environment, and these parameters are adjusted by regulating the temperature and humidity control modules.

[0067] Preferably, the present invention also combines environmental interference parameters, atmospheric transmission parameters and surface reflection parameters to compensate and correct the simulated light, so as to obtain a final and complete dynamic light source parameter library output based on the actual use scenario and the specific environment.

[0068] Specifically, the environmental interference parameters include temperature, humidity, precipitation intensity, and pollutant concentration; the atmospheric transport parameters include attenuation coefficient and visibility, wherein the attenuation coefficient characterizes distance-related brightness attenuation; and the surface reflection parameters include the effects of reflections from surfaces such as ice, different road materials, and vehicle paint on light.

[0069] The final dynamic light source parameter library includes the light source parameters of simulated light under various usage environments.

[0070] S12: Determination of dynamic light source parameters under preliminary test conditions;

[0071] Research on common faults reveals that some electronic streaming rearview mirrors exhibit a washed-out image in bright light areas due to excessive reflectivity of the lens coating or an overly high highlight clipping threshold in the processing algorithm. Others suffer from ghosting due to insufficient lens barrel extinction treatment causing multiple reflections within the lens. In some scenarios, color distortion occurs because the built-in algorithm incorrectly identifies the light source. Therefore, this invention determines the test environment based on the fault type and employs different dynamic light tests for different faults to improve the test coverage of faults.

[0072] Based on this, the present invention constructs a test environment based on fault orientation, specifically including the following steps:

[0073] a. Obtain historical fault detection data and classify the faults into N categories based on the fault symptoms;

[0074] The fault phenomena include screen whitening, motion blur, color distortion, stuttering, screen noise, ghosting, etc.

[0075] b. Extract the light source parameters for each type of fault, including the light source optical parameters, motion parameters, and environmental parameters;

[0076] c. Select the parameters of the dynamic light source for testing and build a preliminary test environment, including a basic test library;

[0077] The basic test library uses dynamic light source parameters determined based on common fault types as the initial test environment. The selection of test light is relatively balanced and does not consider personalized needs.

[0078] S2: Obtain the test results under the initial test environment and the test results of the same batch of products, and extract features;

[0079] S21: Collect image data of the electronic streaming rearview mirror during the test using a camera to obtain test result image data under different dynamic light source tests;

[0080] S22: Extract image features from image data;

[0081] Specifically, the image features include optical performance features, color characteristic features, geometric characteristic features, texture characteristic features, and temporal characteristic features;

[0082] The optical performance characteristics include global average brightness, highlight brightness, shadow detail, dynamic range, glare suppression rate, halo area ratio, and contrast loss rate.

[0083] The color characteristics include average color difference, maximum color difference, white balance error, and color saturation.

[0084] The geometric characteristics include barrel distortion rate, pincushion distortion rate, edge resolution loss, straight line curvature, distance calibration error, and angle retention.

[0085] The texture characteristics include edge sharpness, signal-to-noise ratio, and noise power spectrum;

[0086] The temporal characteristics include light adaptation time, motion blur value, inter-frame consistency, and maximum latency.

[0087] S3: Based on the test results of the initial test environment and the test results of the same batch of products, conduct supplementary tests to establish an adaptive test environment;

[0088] For products of different models and batches, the causes and phenomena of failures may differ. Furthermore, within the same batch, due to the use of identical materials and processes, a single failure may lead to similar problems in other products. Therefore, an adaptive testing environment needs to be constructed to focus on identifying and addressing detection issues already observed in similar products. This application constructs an adaptive testing environment based on the detection results from the initial testing environment and the detection results of products from the same batch. Specifically, it uses a parallel deep random forest model to determine the dynamic light source parameters and the number of test cases for supplementary testing, thus establishing the adaptive testing environment. The model input consists of the image features of the detection results from the initial testing environment and the detection results of products from the same batch, while the output consists of the dynamic light source parameters for supplementary testing and the number of test cases corresponding to each parameter.

[0089] Specifically, the steps include the following:

[0090] S31: Establish a parallel deep random forest model, such as Figure 2 As shown, the model comprises a three-layer forest. The first layer of the forest inputs the image features of the detection results from the initial test environment and the detection results from the same batch of products. ,in, These are the optical performance characteristics, color characteristics, geometric characteristics, texture characteristics, and temporal characteristics of the i-th sample, respectively.

[0091] Level 1 input:

[0092]

[0093] Output of layer p (p=1,2,3):

[0094]

[0095] Where T is the number of trees in each layer of the forest. Let be the model function for the t-th tree in the p-th layer. For the input of the p-th layer;

[0096] The expression is as follows:

[0097]

[0098] [ ; ] represents the feature splicing operation.

[0099] Furthermore, the loss function of the parallel deep random forest model is determined by combining the physical constraints of the light parameters of natural light and artificial light, as well as the testing cost, and is expressed as follows:

[0100]

[0101] in,

[0102] Prediction error is used to measure the difference between the model's prediction and the actual value. For testing costs; For physical constraints; For regularization terms; , , These are the weighting coefficients.

[0103] Among them, the prediction error term The expression is as follows:

[0104]

[0105] Where K is the number of trees in the parallel random forest; For the first The weights of the trees are typically based on the tree's accuracy on the validation set; Let be the predicted value of the k-th tree for the regression objective; This is the true target value in the regression task.

[0106] The Huber loss function is defined as follows:

[0107]

[0108] The hyperparameter for Huber loss controls the switching threshold between squared error and absolute error.

[0109] Test cost item This is used to control the number of test cases and encourage the use of fewer test samples:

[0110]

[0111] The number of test cases currently selected. This represents the number of benchmark test cases used for normalization.

[0112] Physical constraints This is a penalty term that ensures the light source parameters meet physical constraints. It consists of multiple sub-penalty terms, targeting different types of constraints:

[0113]

[0114] in, This represents the penalty value for the i-th constraint. Specifically, it includes:

[0115] Boundary constraints, such as wavelength range and intensity range;

[0116] For each parameter, if it exceeds the boundary, a penalty is applied:

[0117]

[0118] in:

[0119] For the j-th light source parameter, , Let j represent the minimum and maximum allowed values ​​for the j-th parameter.

[0120] Regularization term This item is used to control model complexity and consists of two parts:

[0121]

[0122] in, The feature weight matrix; The square of the Frobenius norm; Let be the depth of the k-th tree; The weighting coefficient for the tree depth penalty.

[0123] S32: Perform model training; obtain image data and light source parameters of historical faults based on historical fault data, as well as the number of test cases determined manually, as the training database;

[0124] S33: Input the image features of the detection results of the initial test environment and the detection results of the same batch of products, and generate the dynamic light source parameters and the number of test cases for supplementary testing based on the trained model;

[0125] S34: Based on the parameters of the dynamic light source used for supplementary testing and the number of test cases, an adaptive test environment is established to obtain the final dynamic light test environment.

[0126] The dynamic light source test optimization model based on parallel deep random forest has significant advantages: its multi-granularity scanning mechanism can intelligently analyze the complex features of the initial test image, accurately capture its nonlinear correlation with the light source parameters, and overcome the limitations of human experience. Through efficient parallel computing, the model outputs the optimized dynamic light parameters and the corresponding adaptive number of test cases in real time—key parameters are automatically allocated more resources, and tests are dynamically added in low-confidence scenarios, avoiding redundancy or omissions caused by uniform allocation, forming a self-evolving test system of "perception-decision-optimization".

[0127] In this embodiment, a dynamically reconfigurable light source control platform is constructed through a systematic adaptive testing environment, driven by a closed-loop decision-making mechanism based on real-time collected test feedback data. The system intelligently explores light source parameters using multi-dimensional optimization algorithms, generating highly customized light source strategies for the core pain points of different testing scenarios. Each round of testing dynamically adjusts parameter combinations based on historical performance, prioritizing the test light source parameters that are most challenging for the current test object, ensuring that test conditions always focus on the sensitive range most likely to expose potential faults. This closed-loop adaptive mechanism not only avoids the coverage blind spots of fixed light source configurations but also actively induces abnormal responses under critical conditions by continuously applying targeted optical pressure. This maximizes defect detection rate and fault mode identification accuracy while ensuring testing efficiency, establishing a dynamically enhanced optical stress testing system for product quality verification.

[0128] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. An automated detection system for in-vehicle electronic streaming rearview mirrors, characterized in that, include Light source dark box, multi-axis light source array, moving light source track, dimming module, temperature and humidity control module, camera, dynamic light test environment determination module, output module; The multi-axis light source array consists of several independently controllable LEDs, used to simulate natural light and artificial light; The multi-axis light source array is mounted on a moving light source track and can move on the track; The dynamic light test environment determination module includes a preliminary test environment determination module, a detection result processing module, and an adaptive test environment determination module, which are used to determine the preliminary test environment and the adaptive test environment. The dimming module and the temperature and humidity control module adjust the dynamic light source parameters and the temperature and humidity of the test environment based on the output of the dynamic light test environment determination module. The camera is used to capture electronic streaming rearview mirror images during testing; The output module outputs the total test result based on the test results under the test environment determined by the dynamic light test environment determination module.

2. The automated detection system for in-vehicle electronic streaming rearview mirrors according to claim 1, characterized in that, The dynamic light source parameters include the light source's optical parameters, motion parameters, and environmental parameters.

3. The automated detection system for in-vehicle electronic streaming rearview mirrors according to claim 2, characterized in that, The optical parameters of the light source include the spectral characteristics and polarization characteristics of the light source; The motion parameters include brightness change parameters, flashing frequency, motion trajectory, moving speed, acceleration, and angle change rate; The environmental parameters are used for environmental coupling of simulated light, and are adjusted by regulating the temperature and humidity control module. 。 4. The automated detection system for in-vehicle electronic streaming rearview mirrors according to claim 1, characterized in that, The dynamic light testing environment determination module specifically includes the following steps: S1: Building the initial test environment; The construction of the preliminary test environment includes the determination of the dynamic light source parameter library and the determination of the dynamic light source parameters in the preliminary test environment; S2: Obtain the test results under the initial test environment and the test results of the same batch of products, and extract features; S3: Based on the test results of the initial test environment and the test results of the same batch of products, conduct supplementary tests to establish an adaptive test environment.

5. The automated detection system for in-vehicle electronic streaming rearview mirrors according to claim 4, characterized in that, The determination of the dynamic light source parameter library includes compensating and correcting the simulated light by combining environmental interference parameters, atmospheric transmission parameters, and surface reflection parameters, to obtain the final and complete dynamic light source parameter library output based on the actual use scenario and the specific environment.

6. The automated detection system for in-vehicle electronic streaming rearview mirrors according to claim 4, characterized in that, The determination of dynamic light source parameters in the preliminary test environment is based on fault-oriented methods, including acquiring historical fault detection data and classifying faults into N categories according to fault phenomena; extracting light source parameters for each category of fault, including light source optical parameters, motion parameters, and environmental parameters. Based on the fault category, select the dynamic light source parameters for testing and construct the initial test environment.

7. The automated detection system for in-vehicle electronic streaming rearview mirrors according to claim 4, characterized in that, The supplementary testing based on the initial test environment results and the test results of the same batch of products, to establish an adaptive test environment, includes: S31: Establish a parallel deep random forest model; S32: Perform model training; S33: Input the image features of the detection results of the initial test environment and the detection results of the same batch of products, and generate the dynamic light source parameters and the number of test cases for supplementary testing based on the trained model; S34: Based on the parameters of the dynamic light source used for supplementary testing and the number of test cases, an adaptive test environment is established to obtain the final dynamic light test environment.

8. The automated detection system for in-vehicle electronic streaming rearview mirrors according to claim 7, characterized in that, The parallel deep random forest model comprises three layers. The first layer of the forest takes into account the image features of the detection results from the initial test environment and the detection results from the same batch of products. ,in, These are the optical performance characteristics, color characteristics, geometric characteristics, texture characteristics, and temporal characteristics of the i-th sample, respectively.

9. The automated detection system for in-vehicle electronic streaming rearview mirrors according to claim 8, characterized in that, The input to the first layer of the parallel deep random forest model: Output of layer p (p=1,2,3): Where T is the number of trees in each layer of the forest. Let be the model function for the t-th tree in the p-th layer. For the input of the p-th layer; The expression is as follows: [ ; ] represents the feature splicing operation.

10. The automated detection system for in-vehicle electronic streaming rearview mirrors according to claim 9, characterized in that, The loss function expression for the parallel deep random forest model is as follows: in, Prediction error is used to measure the difference between the model's prediction and the actual value. For testing costs; For physical constraints; For regularization terms; , , These are the weighting coefficients.

Images