LED dead pixel detection method, device and equipment based on laser reflection and medium

Abstract

The invention discloses an LED dead pixel detection method, device and equipment based on laser reflection and a medium, and relates to the technical field of LED lamp detection, and the method comprises the steps: controlling a laser light source to irradiate a to-be-detected LED lamp; collecting a reflected light signal; preprocessing the reflected light signal to obtain an input signal; and judging whether the LED lamp is normal based on the input signal. According to the invention, laser reflection detection is adopted to replace traditional optical imaging, the monochromaticity and directivity of laser are utilized to overcome ambient light interference, and a darkroom or complex filtering is not needed. Meanwhile, the combination of the laser light source and the photoelectric sensor significantly reduces the cost of the system, and avoids the use of a high-resolution camera. The laser can focus a micron-sized light spot, and the sub-pixel-level micro structure abnormity can be identified with high precision by analyzing the energy change of a reflected light signal. More importantly, according to the method, non-luminous defects such as packaging cracks and colloid stripping can be synchronously detected by capturing the reflection characteristics of the physical structure of the LED instead of depending on the luminous state of the LED, so that the defect detection accuracy is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of LED lamp detection, and particularly relates to an LED bad point detection method, device, equipment and medium based on laser reflection. BACKGROUND

[0002] In the production and quality control process of light emitting diode (LED) devices, bad point detection is a key link to ensure the performance and reliability of the final product. Bad points mainly manifest as complete non-emission (dead points), abnormal brightness (such as dark spots, bright spots), color deviation or physical structure damage (such as package cracks), and other defect morphologies. For a long time, the traditional technical scheme based on optical imaging has been widely used in this detection task. The core of this scheme is to use an industrial camera to collect optical images of the working state of an LED array, and then analyze the brightness and color information of each region in the image through an image processing algorithm, and identify the bad points with abnormal brightness (such as too bright, too dark or not bright) according to the preset discrimination standard.

[0003] However, this traditional detection method relying on external optical imaging faces significant technical challenges in actual application, which seriously affects the accuracy and universality of the detection. One major problem is that it is severely disturbed by environmental light conditions. Stray light can introduce significant background noise, which interferes with the judgment of the true light-emitting state of the LED, resulting in frequent misjudgment or missed detection. To overcome this interference, it is usually necessary to operate in a darkroom environment or rely on complex real-time filtering and background light suppression algorithms. The former increases the site requirements and operation complexity, and the latter significantly increases the computational burden and implementation difficulty of the system.

[0004] Further, to achieve effective detection of high-density, miniaturized LED arrays, especially to identify micro-defects, it is often necessary to configure an optical imaging system with high spatial resolution, including high-performance lenses and image sensors. This dependence on high-resolution hardware devices directly leads to a significant increase in system construction and maintenance costs, making the overall detection scheme less economical. What is particularly critical is that, limited by the inherent physical resolution of the optical imaging system and the optical diffraction limit, it is often difficult for traditional methods to clearly distinguish and accurately identify micro-bad points or subtle brightness abnormalities (i.e., sub-pixel level defects) with a size smaller than a single imaging pixel. Such micro-defects information is easily lost or overwhelmed by noise during image acquisition, making it difficult to meet the increasingly high quality requirements.

[0005] In addition, the core mechanism of the traditional optical imaging method relies on capturing and analyzing the light-emitting characteristics of the LED. Therefore, its detection capability has a fundamental limitation: for non-light-emitting physical defects that do not affect the current light-emitting state, such as fine cracks inside the packaging material, internal connection problems that have not yet caused obvious electrical or optical failure, or scratches on the surface of the package, etc., the method is basically ineffective in identifying. These "implicit" defects may not show functional abnormalities in initial detection, but they may expand due to stress, temperature changes, or aging factors during subsequent use, eventually evolving into functional failure points, causing potential quality risks and reliability problems.

[0006] In summary, the traditional LED defect detection method based on optical imaging has obvious shortcomings in dealing with environmental light interference, controlling detection cost, improving micro-defect recognition accuracy, and covering non-light-emitting physical defects. Its sensitivity to environmental conditions, high system cost, limitations in detecting sub-pixel level micro-defects, and detection blind spots for non-light-emitting defects collectively restrict the wide application range of the technology and the accuracy of the final detection results. These inherent limitations highlight the urgent need for a new detection technology solution that is more efficient, more robust, cost-controllable, and can comprehensively cover all types of defects. SUMMARY

[0007] Embodiments of the present application provide a laser reflection-based LED defect detection method, device, equipment and medium, aiming to solve the problem of high cost and low accuracy of traditional LED defect detection.

[0008] In a first aspect, embodiments of the present application provide a laser reflection-based LED defect detection method, which includes:

[0009] controlling a laser light source to irradiate laser light to a to-be-tested LED lamp;

[0010] acquiring a reflected light signal reflected by the to-be-tested LED lamp;

[0011] preprocessing the reflected light signal to obtain an input signal;

[0012] judging whether the to-be-tested LED lamp is a normal LED lamp based on the input signal.

[0013] A further technical solution is that the controlling a laser light source to irradiate laser light to a to-be-tested LED lamp includes:

[0014] acquiring a preset PWM signal;

[0015] controlling the laser light source to irradiate laser light to the to-be-tested LED lamp based on the PWM signal.

[0016] A further technical solution is that the preprocessing the reflected light signal to obtain an input signal includes:

[0017] The reflected light signal is subjected to noise reduction processing to obtain the input signal, and the noise reduction processing includes wavelet transform denoising.

[0018] Further technical solutions are that the determining whether the to-be-tested LED lamp is a normal LED lamp based on the input signal includes:

[0019] Obtaining a signal energy of the input signal when the to-be-tested LED lamp is in an off state to obtain a static energy value;

[0020] Obtaining a signal energy of the input signal at a moment when the to-be-tested LED lamp is turned on to obtain a dynamic energy value;

[0021] Obtaining a ratio of the dynamic energy value to the static energy value;

[0022] Determining whether the ratio is greater than a preset ratio threshold value;

[0023] If the ratio is greater than the preset ratio threshold value, determining that the to-be-tested LED lamp is a normal LED lamp.

[0024] Further technical solutions are that the obtaining the signal energy of the input signal when the to-be-tested LED lamp is in the off state to obtain the static energy value includes:

[0025] Obtaining signal energies of multiple sampling points of the input signal when the to-be-tested LED lamp is in the off state;

[0026] Calculating an average value of the signal energies of the multiple sampling points to obtain the static energy value.

[0027] Further technical solutions are that the obtaining the signal energy of the input signal at the moment when the to-be-tested LED lamp is turned on to obtain the dynamic energy value includes:

[0028] Determining a target sampling point corresponding to the moment when the to-be-tested LED lamp is turned on;

[0029] Obtaining a signal energy of the input signal at the target sampling point to obtain the dynamic energy value.

[0030] Further technical solutions are that the determining whether the to-be-tested LED lamp is a normal LED lamp based on the input signal includes:

[0031] Performing feature extraction on the input signal to obtain an input feature;

[0032] Determining whether the to-be-tested LED lamp is a normal LED lamp according to the input feature through a pre-trained machine learning model.

[0033] In a second aspect, the embodiments of the present application also provide a laser reflection-based LED dead pixel detection device, which comprises units for executing the above method.

[0034] In a third aspect, the embodiments of the present application also provide a computer device, which comprises a memory and a processor, the memory stores a computer program, and the processor implements the above method when executing the computer program.

[0035] In a fourth aspect, the embodiments of the present application also provide a computer readable storage medium, which stores a computer program, and the computer program can implement the above method when executed by a processor.

[0036] The embodiments of the present application provide a laser reflection-based LED dead pixel detection method, device, equipment and medium. The method comprises: controlling a laser light source to irradiate a laser to a to-be-tested LED lamp; collecting a reflected light signal reflected by the to-be-tested LED lamp; pre-processing the reflected light signal to obtain an input signal; and judging whether the to-be-tested LED lamp is a normal LED lamp based on the input signal. The present application replaces the traditional optical imaging with a laser reflection detection mechanism, which firstly avoids the problem of environmental light interference. The monochromaticity and directionality of the laser enable it to stably irradiate the surface of the to-be-tested LED under complex lighting conditions, and the physical information carried by the reflected light signal is not polluted by environmental stray light. Secondly, the combination of the laser light source and the photoelectric sensor significantly reduces the system cost, and there is no need for a high-resolution camera and a supporting optical module. In terms of detection accuracy, the laser beam can be focused to a micron-level light spot, and through the analysis of the energy change of the reflected light signal, sub-pixel-level micro-structure abnormalities (such as electrode micro-fracture) can be identified. Most importantly, this technology can detect non-luminous defects such as packaging cracks and colloid peeling by capturing the reflection characteristics of the physical structure of the LED lamp bead (rather than relying on the luminous state). The entire detection process forms a closed loop: laser irradiation excitation physical feedback -> reflected signal acquisition -> pre-processing noise reduction -> intelligent decision, which finally realizes the dimensional upgrade from single luminous detection to physical structure integrity detection, and systematically solves the four bottlenecks of environmental dependence, cost pressure, micro-defect missed detection and non-luminous defect blind area of the traditional method. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0038] Figure 1 A flowchart of a laser reflection-based LED dead pixel detection method provided by the embodiments of the present application is shown in the figure.

[0039] Figure 2 A laser light path diagram of an LED bad point detection method based on laser reflection provided for an embodiment of the present application;

[0040] Figure 3 A schematic diagram of a laser scanning a to-be-tested LED lamp provided for an embodiment of the present application;

[0041] Figure 4 A schematic block diagram of a computer device provided for an embodiment of the present application. DETAILED DESCRIPTION

[0042] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0043] It should be understood that, when used in the specification and the appended claims, the terms "comprise" and "include" indicate the presence of described features, integers, steps, operations, elements, and / or components, but do not exclude one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0044] It should also be understood that the terms used in the present application specification are only for the purpose of describing particular embodiments and are not intended to limit the present application. As used in the present application specification and the appended claims, the singular forms "a", "an" and "the" are intended to include the plural forms, unless the context clearly indicates otherwise.

[0045] It should be further understood that the term "and / or" used in the present application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0046] As used in the present application specification and the appended claims, the term "if" can be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if [a described condition or event] is detected" can be interpreted to mean "upon determining" or "in response to determining" or "upon detecting [a described condition or event]" or "in response to detecting [a described condition or event]", depending on the context.

[0047] Please refer to Figure 1The embodiment of the present application provides a LED bad point detection method based on laser reflection, which comprises the following steps:

[0048] S1, controlling a laser light source to irradiate laser to a to-be-detected LED lamp.

[0049] In specific implementation, the laser light source can be preconfigured by a person skilled in the art, and the present application is not specifically limited. The laser light source can be a laser generator. The to-be-detected LED lamp refers to an LED lamp to be detected whether there is a fault.

[0050] In the present application, the laser light source is controlled to irradiate laser to the to-be-detected LED lamp, and the irradiation angle can be 45-90°, which is not specifically limited in the present application. Specifically, the laser light path diagram is as shown in the figure. Figure 2

[0051] In some preferred embodiments, the above step "controlling the laser light source to irradiate laser to the to-be-detected LED lamp" specifically comprises the following steps: acquiring a preset PWM signal; and controlling the laser light source to irradiate laser to the to-be-detected LED lamp based on the PWM signal.

[0052] In specific implementation, the brightness of the laser light source can be regulated through the PWM signal, different laser brightness can be adopted for different materials, and the regulation of the laser brightness is realized by adjusting the duty ratio of the PWM signal, so that the LED lamp of different materials can be adapted.

[0053] S2, collecting reflected light signals reflected by the to-be-detected LED lamp.

[0054] In specific implementation, the reflected light signals reflected by the to-be-detected LED lamp are collected through a sensor. The sensor can be a photodiode, a CCD or CMOS camera, a high-sensitivity photoelectric sensor (such as an APD avalanche diode), and is matched with a narrow-band optical filter (±5nm bandwidth); a receiver is coaxially offset with a laser, and a mixed signal of mirror reflection and diffuse reflection is captured.

[0055] S3, pre-processing the reflected light signals to obtain input signals.

[0056] In specific implementation, the reflected light signals are pre-processed to obtain input signals. Through pre-processing, the accuracy of input data can be improved, so that the accuracy of subsequent calculation is improved.

[0057] For example, in some preferred embodiments, the above step "pre-processing the reflected light signals to obtain input signals" specifically comprises the following steps: performing noise reduction processing on the reflected light signals to obtain the input signals, and the noise reduction processing comprises wavelet transform denoising.

[0058] ​In specific implementation, the reflected light signal is subjected to noise reduction processing to obtain the input signal, and the noise reduction processing includes wavelet transform denoising. Through the noise reduction processing, the clutter in the reflected light signal can be filtered out, thereby reducing noise interference and improving the accuracy of subsequent calculation.

[0059] In the embodiments of the present application, the application of wavelet transform denoising solves the problem of optimizing the signal-to-noise ratio of the laser reflection signal. Due to the complex scattering of laser in the multi-layer medium (semiconductor / encapsulation glue / fluorescent powder) in the LED, the reflection signal is easily disturbed by circuit electromagnetic noise and mechanical vibration, and presents non-stationary characteristics. The wavelet transform performs joint analysis in time-frequency domain, effectively filters out fixed frequency band noise while retaining the mutation characteristics of the signal (such as reflection pulses caused by cracks). The multi-scale decomposition characteristics are particularly suitable for separating the weak high-frequency response (such as the scattering characteristics of micron-level cracks) of sub-pixel-level defects from low-frequency background noise. After the input signal is preprocessed, the signal-to-noise ratio of the effective features is significantly improved, so that the subsequent decision module can more sensitively identify energy threshold changes or machine learning features, especially ensuring the detection lower limit of small defects. This processing breaks through the contradiction between retaining defect characteristics and suppressing noise in traditional filtering algorithms.

[0060] S4, judging whether the to-be-tested LED lamp is a normal LED lamp based on the input signal.

[0061] In specific implementation, whether the to-be-tested LED lamp is a normal LED lamp is judged based on the input signal.

[0062] The specific judgment principle is that a normally working LED will reflect laser, and due to its light-emitting characteristics, the intensity and distribution of the reflected light will be different from that of a bad point. A bad point (such as a dead lamp or a dim lamp) will change the characteristics of the reflected light due to abnormal light emission or non-light emission. Signal processing: the reflected light signal is captured by a sensor (such as a photodetector or a camera), and the intensity, distribution or spectral characteristics of the reflected light are analyzed by an algorithm to determine whether the LED is normal.

[0063] For example, in some preferred embodiments, the above step of "judging whether the to-be-tested LED lamp is a normal LED lamp based on the input signal" specifically includes the following steps: obtaining the signal energy of the input signal when the to-be-tested LED lamp is in an off state to obtain a static energy value; obtaining the signal energy of the input signal at the instant when the to-be-tested LED lamp is turned on to obtain a dynamic energy value; obtaining the ratio of the dynamic energy value to the static energy value; judging whether the ratio is greater than a preset ratio threshold; and if the ratio is greater than the preset ratio threshold, determining that the to-be-tested LED lamp is a normal LED lamp.

[0064] In the embodiment, during the testing process, first, the to-be-tested LED lamp is turned off, the signal energy of the input signal when the to-be-tested LED lamp is in the off state is obtained, and a static energy value is obtained.

[0065] Further, the to-be-tested LED lamp is turned on, the signal energy of the input signal at the instant when the to-be-tested LED lamp is turned on is obtained, and a dynamic energy value is obtained.

[0066] Further, a ratio of the dynamic energy value to the static energy value is obtained; it is determined whether the ratio is greater than a preset ratio threshold value; if the ratio is greater than the preset ratio threshold value, it is determined that the to-be-tested LED lamp is a normal LED lamp. The ratio threshold value can be set by a person skilled in the art, and the present application is not specifically limited, for example, set to 1.5.

[0067] Further, if the ratio is greater than the preset ratio threshold value, it is determined that the to-be-tested LED lamp is a normal LED lamp.

[0068] In the embodiment of the present application, the static energy (E_static) measured when the to-be-tested LED lamp is turned off represents the reflection characteristic of the base material; the dynamic energy (E_dynamic) at the instant when the to-be-tested LED lamp is turned on includes the reflection of the base and the working vibration component of the to-be-tested LED lamp (an additional laser amount generated due to current jitter at the instant when the LED lamp is turned on). The normal LED makes R>1.5 (for example, R=2.1±0.3) due to the vibration component, and the dead lamp makes R≈1 due to no vibration. Thus, reliable distinction between the normal LED lamp and the abnormal LED lamp can be achieved.

[0069] In some preferred embodiments, the above step of "obtaining the signal energy of the input signal when the to-be-tested LED lamp is in the off state, and obtaining a static energy value" specifically includes the following steps: obtaining the signal energy of a plurality of sampling points of the input signal when the to-be-tested LED lamp is in the off state; and calculating the average value of the signal energy of the plurality of sampling points, and obtaining the static energy value.

[0070] In the embodiment, the number of sampling points can be set by a person skilled in the art, for example, set to 60, and the present application is not specifically limited. By calculating the average value of the plurality of sampling points as the static energy value, the accuracy can be effectively improved, and the error caused by single-point data fluctuation can be avoided.

[0071] It should be noted that the signal energy of each sampling point = in-phase signal square (I 2 ) + quadrature signal square (Q 2 )

[0072] In the embodiment, the multi-point average processing of the static energy value further improves the stability of the base signal. Since the laser may be diffusely reflected on the LED surface, single-point sampling is easily disturbed by slight position deviation or surface contamination. By collecting multiple consecutive sampling points in the off state and calculating the average value, the random noise (such as dust scattering) and the instantaneous fluctuation caused by mechanical vibration can be effectively suppressed. This operation is equivalent to constructing a statistical model of the static background, making the characterization of the base reflection energy closer to the intrinsic properties of the material. The static value processed as the denominator of the energy ratio can avoid distortion of the ratio caused by single-point abnormalities, especially ensuring the identification specificity of weak defects (such as surface cracks) and preventing misjudgment.

[0073] In some preferred embodiments, the above step of "obtaining the signal energy of the input signal at the opening moment of the LED lamp to be tested to obtain a dynamic energy value" specifically includes the following steps: determining a target sampling point corresponding to the opening moment of the LED lamp to be tested; and obtaining the signal energy of the input signal at the target sampling point to obtain the dynamic energy value.

[0074] In specific implementation, the time point corresponding to the opening moment of the LED lamp to be tested is obtained, the sampling point corresponding to the time point is taken as the target sampling point, the signal energy of the input signal at the target sampling point is obtained, and the dynamic energy value is obtained, so that the signal energy of the input signal at the opening moment of the LED lamp to be tested can be accurately collected.

[0075] In the embodiment, the dynamic energy value is taken from a single sampling point at the opening moment of the LED, and the technical effect lies in the sensitivity of capturing the transient response. The characteristic vibration is generated at the moment (microsecond level) of normal LED power-on, and the response is delayed or missing for a bad point. By accurately synchronizing the laser pulse with the LED drive signal (time difference <1 μs), the transient energy peak is captured at the target sampling point.

[0076] In some preferred embodiments, the above step of "judging whether the LED lamp to be tested is a normal LED lamp based on the input signal" specifically includes the following steps: extracting input features from the input signal; and judging whether the LED lamp to be tested is a normal LED lamp according to the input features by a pre-trained machine learning model.

[0077] In specific implementation, the machine learning model is trained in advance by labeled training data, so that the machine learning model has the ability to identify whether the LED lamp to be tested is normal based on the input features. The machine learning model can be specifically an SVM classifier, which is not specifically limited by the present application.

[0078] In specific application, the input features are obtained by feature extraction of the input signal, and the input features can specifically include reflected light intensity, phase shift amount, and spectral features, which are not specifically limited by the present application.

[0079] Further, the input features are input to the machine learning model, and the machine learning model determines the category of the to-be-tested LED lamp based on the input features, the category including normal and abnormal. If the category is normal, it indicates that the to-be-tested LED lamp is a normal LED lamp, and if the category is abnormal, it indicates that the to-be-tested LED lamp is an abnormal LED lamp.

[0080] In the embodiment of the present application, the introduction of the machine learning model realizes the fusion decision of multi-dimensional defect features. By deeply extracting the reflected light intensity, phase offset and spectral features of the preprocessed reflected signal, a high-dimensional defect representation space beyond the simple energy ratio can be constructed. The model learns the implicit patterns of normal and defect samples through training. This nonlinear decision boundary breaks through the limitations of traditional threshold methods, can simultaneously identify complex defects (such as cracks + virtual welding) and quantify the severity of defects, and significantly improves the generalization ability and adaptability of the detection system.

[0081] Further, the above two ways of determining whether the to-be-tested LED lamp is abnormal can be cross-validated with each other, thereby improving the accuracy of the determination.

[0082] The embodiment of the present application proposes a laser reflection-based LED dead point detection method, which comprises: controlling a laser light source to irradiate laser light to a to-be-tested LED lamp; collecting a reflected light signal reflected by the to-be-tested LED lamp; preprocessing the reflected light signal to obtain an input signal; and determining whether the to-be-tested LED lamp is a normal LED lamp based on the input signal. The present application replaces the traditional optical imaging with a laser reflection detection mechanism, which first avoids the problem of environmental light interference. The monochromaticity and directionality of laser light enable it to stably irradiate the surface of the to-be-tested LED under complex lighting conditions, and the physical information carried by the reflected light signal is not polluted by environmental stray light. Secondly, the combination of the laser light source and the photoelectric sensor significantly reduces the system cost, without the need for a high-resolution camera and a supporting optical module. In terms of detection accuracy, the laser beam can be focused to a micron-level spot, and by analyzing the energy change of the reflected light signal, sub-pixel-level micro-structure abnormalities (such as electrode micro-fracture) can be identified. Most importantly, this technology can simultaneously detect non-luminous defects such as package cracks and colloid peeling by capturing the reflection characteristics of the physical structure of the LED lamp bead (rather than relying on the luminous state). The entire detection process forms a closed loop: laser irradiation excitation physical feedback -> reflected light signal acquisition -> preprocessing noise reduction -> intelligent decision, ultimately realizing the dimensional upgrade from single luminous detection to physical structure integrity detection, and systematically solving the four major bottlenecks of environmental dependence, cost pressure, micro-defect missed detection and non-luminous defect blind area of traditional methods.

[0083] The method of the embodiment of the present application is applied to an LED defect detection system, and the LED defect detection system comprises:

[0084] Laser light source: Provides a stable laser beam, usually selecting visible light lasers (such as red or green light).

[0085] Optical system: Includes lenses, mirrors, etc., for focusing the laser beam and guiding the reflected light.

[0086] Sensor: Used to capture reflected light signals, can be photodiodes, CCD or CMOS cameras.

[0087] Signal processing unit: Analyzes the reflected light signal, extracts features and judges the LED state.

[0088] Motion control platform: Used to move the laser head or LED display screen to achieve full-screen scanning, as shown in Figure 3 .

[0089] Further, the detection step includes:

[0090] Laser scanning: Laser beam scans LED display screen point by point or line by line.

[0091] Reflection light collection: Sensor captures reflected light signal of each LED in real time.

[0092] Signal analysis: Normal LED: Higher reflectivity and uniform distribution. Bad point LED: Lower reflectivity or abnormal distribution.

[0093] Bad point positioning: According to the analysis results, mark the position of bad points.

[0094] Data output: Generate detection report, display bad point position and quantity.

[0095] Further, the core components are described in detail as follows:

[0096] Laser emission module:

[0097] Use wavelength-specific wavelength modulation laser to emit frequency pulse laser to avoid environmental light interference;

[0098] Laser incidence angle is set to 45-90° to optimize reflected signal intensity;

[0099] Optional multi-beam array design, using spatial light modulator principle, through liquid crystal, acousto-optic or electro-optic modulator dynamically adjusting laser wavefront phase, splitting single beam into multiple beams. Use computer-generated hologram to modulate light field, generate any shape of multi-beam (such as optical tweezers, optical micro-operation). Real-time adjustment of beam number, direction and intensity, covering the detection area.

[0100] Reflection signal receiving module: used to capture reflected light signals, which can be photodiodes, CCD or CMOS cameras, high-sensitivity photoelectric sensors (such as APD avalanche diodes), combined with narrow-band optical filters (±5nm bandwidth); the receiver is coaxially offset from the laser, capturing mixed signals of specular and diffuse reflection.

[0101] Motion control module:

[0102] Two-dimensional precision guide rail carries detection head, driven by stepper motor, positioning accuracy ±0.1mm;

[0103] Adaptive scanning path planning, supporting curved LED detection.

[0104] Signal processing unit:

[0105] Real-time signal processing based on FPGA, extracting reflected light intensity, phase shift and spectral features;

[0106] Wavelet transform is used for denoising, combined with machine learning models (such as SVM classifier) to distinguish normal / abnormal signals. The following is the detailed implementation process and key technical points of SVM:

[0107] I. Feature selection method

[0108] The goal of feature selection is to remove redundant or irrelevant features, improve model generalization ability and computational efficiency.

[0109] Automatically select features during model training.

[0110] L1 regularization (LASSO): Using L1 regularization in SVM (such as `penalty='l 1'` for linear SVM) makes some feature coefficients tend to zero, achieving feature sparsification.

[0111] Feature importance based on tree models: Use random forests or XGBoost to obtain feature importance after training, and select Top-N features to input SVM.

[0112] II. Model optimization process

[0113] The core optimization goal of SVM is to maximize the classification margin and minimize the generalization error by adjusting the hyperparameters and kernel functions.

[0114] 1. Data preprocessing:

[0115] Standardization: SVM is sensitive to feature scale, so data needs to be standardized (such as Z-score normalization).

[0116] Handling class imbalance: If the data distribution is imbalanced, use the `class_weight` parameter to adjust class weights (e.g., `class_weight='balanced'`).

[0117] 2. Kernel function selection:

[0118] Select a kernel function based on data characteristics.

[0119] 3. Hyperparameter tuning:

[0120] Key parameters:

[0121] C (regularization parameter): Controls the trade-off between margin and misclassification.

[0122] Small C: Large margin, allows more misclassifications (prevents overfitting).

[0123] Large C: Small margin, reduces misclassifications (may overfit).

[0124] Gamma (RBF kernel parameter): Controls the influence range of individual samples.

[0125] Large gamma: Complex model, may overfit.

[0126] Small gamma: Simple model, may underfit. Tuning method: Grid Search: Traverse parameter combinations, select optimal values through cross-validation.

[0127] 4. Model evaluation and validation

[0128] Cross-validation: Use K-fold cross-validation (e.g., 5-fold) to avoid overfitting.

[0129] Evaluation metrics: Balanced data: Accuracy.

[0130] Imbalanced data: F1-score, AUC-ROC curve.

[0131] Learning curve analysis: Observe the change of training set and validation set errors with the number of samples to determine whether underfitting or overfitting occurs.

[0132] Further, the present application sets up adaptive environmental compensation:

[0133] Set up a reference laser path to real-time calibrate the influence of environmental temperature and humidity on the optical path.

[0134] I. Correction method:

[0135] 1. Use a standard light intensity meter (such as a lux meter) as a reference to record sensor output at multiple known light points (such as 10 Lux, 100 Lux, 1000 Lux).

[0136] 2. Fit a linear or polynomial curve to establish the mapping between sensor output and true Lux.

[0137] 3. Temperature and humidity compensation: Sensor output drifts with temperature and humidity (e.g., photodiode sensitivity changes). Integrate a temperature and humidity sensor to monitor temperature and humidity in real-time. Dynamically adjust readings based on temperature and humidity-output characteristic table.

[0138] 4. Dynamic response parameter adjustment:

[0139] Optimize the response speed and stability of the system to sudden changes in light.

[0140] Response time: Adjust the filter time constant to avoid rapid jitter or delay.

[0141] Hysteresis threshold: Set a light change threshold to prevent frequent switching (e.g., ±10% Lux change to trigger adjustment).

[0142] II. System-level verification and optimization

[0143] 1. Static scene verification

[0144] Method: Run the system under fixed light and check the output stability. Indicators: Brightness fluctuation range (e.g., ±2%), sensor reading standard deviation.

[0145] 2. Dynamic scene test:

[0146] Simulate sudden light changes: Use adjustable light sources (e.g., LED lights + PWM controller) to simulate rapid light intensity changes.

[0147] Verify response: Rise time: time required to reach 90% of the target value from dark to light (e.g., <200ms).

[0148] Manual fine-tuning: Allow users to override automatic brightness and record preferences (e.g., "darker / brighter").

[0149] Self-learning algorithm: Dynamically update the target brightness curve based on multiple user manual adjustments.

[0150] Further, in the present invention, multi-physical field coupling analysis is performed:

[0151] Combined with the thermal imaging module (optional), synchronously monitor the temperature anomalies of the LED micro area, and improve the detection rate of complex defects. In thermal imaging technology, the coupling of temperature signal data and bad point judgment data is mainly used to eliminate abnormal data (i.e., "bad points") caused by pixel failure, noise interference, or hardware defects in infrared detectors.

[0152] Coupling method of temperature signal and bad point data:

[0153] (1)Bad pixel marking and data rejection, static and dynamic bad pixel positions are stored as a binary mask (e.g. marked as 1 in the matrix indicates a bad pixel, 0 indicates normal). In the temperature signal processing flow, directly reject the bad pixel data marked by the mask to avoid its participation in subsequent calculations.

[0154] (2) Bad pixel compensation and data repair:

[0155] Using the temperature values of the normal pixels around the bad pixel, estimate the temperature of the bad pixel position through interpolation algorithm (such as bilinear interpolation, median filter).

[0156] Example: If a pixel is marked as a bad pixel, replace it with the average value of its 8 adjacent pixels. For dynamic bad pixels, predict their reasonable values based on historical temperature data or environmental parameters (such as time series prediction model).

[0157] (3) Bad pixel processing in non-uniformity correction (NUC)

[0158] Calibration and bad pixel coupling: In the NUC process, when calibrating the detector response with a blackbody radiation source, identify bad pixels simultaneously and update the mask. Exclude bad pixels in the correction coefficient calculation to prevent their impact on overall non-uniformity compensation.

[0159] Adaptive correction: Dynamic bad pixels may cause traditional NUC to fail, so real-time bad pixel judgment data needs to be combined to dynamically adjust the correction parameters.

[0160] Coupling of temperature signal data and bad pixel judgment data, the essence is to eliminate the influence of detector hardware defects on temperature measurement accuracy through hardware calibration, real-time detection and algorithm compensation.

[0161] Its technical core lies in:

[0162] 1. Accurate determination of bad pixels (static and dynamic);

[0163] 2. Efficient repair strategy (rejection, interpolation or redundant replacement);

[0164] 3. Seamless integration with signal processing flow (such as NUC, temperature calibration).

[0165] Corresponding to the above LED bad pixel detection method based on laser reflection, the present application also provides a kind of LED bad pixel detection device based on laser reflection. The LED bad pixel detection device based on laser reflection includes unit for executing the above LED bad pixel detection method based on laser reflection, and the LED bad pixel detection device based on laser reflection can be configured in desktop computer, tablet computer, portable computer, etc. terminal. Specifically, the LED bad pixel detection device based on laser reflection includes:

[0166] Irradiation unit for controlling laser light source to irradiate laser to LED lamp to be measured;

[0167] a collection unit, configured to collect a reflected light signal reflected by the to-be-tested LED lamp;

[0168] a preprocessing unit, configured to preprocess the reflected light signal to obtain an input signal;

[0169] a judgment unit, configured to judge whether the to-be-tested LED lamp is a normal LED lamp based on the input signal.

[0170] In some preferred embodiments, the control of the laser light source to irradiate laser light to the to-be-tested LED lamp comprises:

[0171] acquiring a preset PWM signal;

[0172] controlling the laser light source to irradiate laser light to the to-be-tested LED lamp based on the PWM signal.

[0173] In some preferred embodiments, the preprocessing of the reflected light signal to obtain an input signal comprises:

[0174] performing noise reduction processing on the reflected light signal to obtain the input signal, and the noise reduction processing comprises wavelet transform denoising.

[0175] In some preferred embodiments, the judgment of whether the to-be-tested LED lamp is a normal LED lamp based on the input signal comprises:

[0176] acquiring a signal energy of the input signal when the to-be-tested LED lamp is in an off state to obtain a static energy value;

[0177] acquiring a signal energy of the input signal at a moment when the to-be-tested LED lamp is turned on to obtain a dynamic energy value;

[0178] acquiring a ratio of the dynamic energy value to the static energy value;

[0179] judging whether the ratio is greater than a preset ratio threshold value;

[0180] if the ratio is greater than the preset ratio threshold value, determining that the to-be-tested LED lamp is a normal LED lamp.

[0181] In some preferred embodiments, the acquisition of the signal energy of the input signal when the to-be-tested LED lamp is in the off state to obtain the static energy value comprises:

[0182] acquiring signal energies of multiple sampling points of the input signal when the to-be-tested LED lamp is in the off state;

[0183] calculating an average value of the signal energies of the multiple sampling points to obtain the static energy value.

[0184] In some preferred embodiments, the acquiring the signal energy of the input signal at the instant when the LED lamp to be tested is turned on to obtain a dynamic energy value comprises:

[0185] determining a target sampling point corresponding to the instant when the LED lamp to be tested is turned on;

[0186] acquiring the signal energy of the input signal at the target sampling point to obtain the dynamic energy value.

[0187] In some preferred embodiments, the determining whether the LED lamp to be tested is a normal LED lamp based on the input signal comprises:

[0188] performing feature extraction on the input signal to obtain an input feature;

[0189] determining whether the LED lamp to be tested is a normal LED lamp according to the input feature through a pre-trained machine learning model.

[0190] It should be noted that the specific implementation process of the above-mentioned LED bad point detection device based on laser reflection and each unit can be clearly understood by those skilled in the art, which can be referred to the corresponding description in the foregoing method embodiments. For the convenience and brevity of description, it will not be repeated here.

[0191] The above-mentioned LED bad point detection device based on laser reflection can be realized in the form of a computer program, which can run on a computer device as shown in Figure 4 .

[0192] Please refer to Figure 4 , Figure 4 is a schematic block diagram of a computer device provided by an embodiment of the present application. The computer device 500 can be a terminal or a server, wherein the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a personal digital assistant, a wearable device, and an electronic device with a communication function. The server can be a stand-alone server or a server cluster composed of multiple servers.

[0193] The computer device 500 includes a processor 502, a memory, and a network interface 505 connected through a system bus 501, wherein the memory can include a non-volatile storage medium 503 and an internal memory 504.

[0194] The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, the processor 502 can execute an LED bad point detection method based on laser reflection.

[0195] The processor 502 is configured to provide computing and control capabilities to support the operation of the entire computer device 500.

[0196] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503, which, when executed by the processor 502, can cause the processor 502 to execute a laser reflection-based LED bad point detection method.

[0197] The network interface 505 is configured to communicate with other devices via a network. Those skilled in the art can understand that the above structure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device 500 to which the scheme of the present application is applied. The specific computer device 500 can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0198] The processor 502 is configured to run the computer program 5032 stored in the memory to implement the following steps:

[0199] controlling a laser light source to irradiate a laser light to a to-be-tested LED lamp;

[0200] acquiring a reflected light signal reflected by the to-be-tested LED lamp;

[0201] preprocessing the reflected light signal to obtain an input signal;

[0202] judging whether the to-be-tested LED lamp is a normal LED lamp based on the input signal.

[0203] In some preferred embodiments, the step of controlling the laser light source to irradiate the laser light to the to-be-tested LED lamp comprises:

[0204] acquiring a preset PWM signal;

[0205] controlling the laser light source to irradiate the laser light to the to-be-tested LED lamp based on the PWM signal.

[0206] In some preferred embodiments, the step of preprocessing the reflected light signal to obtain the input signal comprises:

[0207] performing noise reduction processing on the reflected light signal to obtain the input signal, and the noise reduction processing comprises wavelet transform denoising.

[0208] In some preferred embodiments, the step of judging whether the to-be-tested LED lamp is a normal LED lamp based on the input signal comprises:

[0209] acquiring a signal energy of the input signal when the to-be-tested LED lamp is in an off state to obtain a static energy value;

[0210] obtaining a signal energy of the input signal at a moment when the to-be-tested LED lamp is turned on, to obtain a dynamic energy value;

[0211] obtaining a ratio of the dynamic energy value and the static energy value;

[0212] determining whether the ratio is greater than a preset ratio threshold value;

[0213] if the ratio is greater than the preset ratio threshold value, determining that the to-be-tested LED lamp is a normal LED lamp.

[0214] In some preferred embodiments, the obtaining of the signal energy of the input signal at the moment when the to-be-tested LED lamp is turned on, to obtain a dynamic energy value, comprises:

[0215] obtaining signal energies of the input signal at multiple sampling points when the to-be-tested LED lamp is turned off;

[0216] calculating an average value of the signal energies of the multiple sampling points, to obtain the static energy value.

[0217] In some preferred embodiments, the obtaining of the signal energy of the input signal at the moment when the to-be-tested LED lamp is turned on, to obtain a dynamic energy value, comprises:

[0218] determining a target sampling point corresponding to the moment when the to-be-tested LED lamp is turned on;

[0219] obtaining a signal energy of the input signal at the target sampling point, to obtain the dynamic energy value.

[0220] In some preferred embodiments, the determining of whether the to-be-tested LED lamp is a normal LED lamp based on the input signal comprises:

[0221] performing feature extraction on the input signal to obtain an input feature;

[0222] determining whether the to-be-tested LED lamp is a normal LED lamp according to the input feature through a pre-trained machine learning model.

[0223] It should be understood that, in the embodiments of the present application, the processor 502 can be a central processing unit (CPU), and the processor 502 can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0224] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the above-mentioned embodiments.

[0225] Therefore, the present application also provides a storage medium. The storage medium can be a computer-readable storage medium. The storage medium stores a computer program. The computer program is executed by a processor to make the processor perform the following steps:

[0226] controlling a laser light source to irradiate laser light to the LED lamp to be measured;

[0227] acquiring a reflected light signal reflected by the LED lamp to be measured;

[0228] preprocessing the reflected light signal to obtain an input signal;

[0229] judging whether the LED lamp to be measured is a normal LED lamp based on the input signal.

[0230] In some preferred embodiments, the controlling the laser light source to irradiate laser light to the LED lamp to be measured comprises:

[0231] acquiring a preset PWM signal;

[0232] controlling the laser light source to irradiate laser light to the LED lamp to be measured based on the PWM signal.

[0233] In some preferred embodiments, the preprocessing the reflected light signal to obtain an input signal comprises:

[0234] The reflected light signal is subjected to noise reduction processing to obtain the input signal, and the noise reduction processing includes wavelet transform denoising.

[0235] In some preferred embodiments, the determining whether the to-be-tested LED lamp is a normal LED lamp based on the input signal comprises:

[0236] Obtaining a signal energy of the input signal when the to-be-tested LED lamp is in an off state to obtain a static energy value;

[0237] Obtaining a signal energy of the input signal at a moment when the to-be-tested LED lamp is turned on to obtain a dynamic energy value;

[0238] Obtaining a ratio of the dynamic energy value to the static energy value;

[0239] Determining whether the ratio is greater than a preset ratio threshold value;

[0240] If the ratio is greater than the preset ratio threshold value, determining that the to-be-tested LED lamp is a normal LED lamp.

[0241] In some preferred embodiments, the obtaining of the signal energy of the input signal when the to-be-tested LED lamp is in the off state to obtain the static energy value comprises:

[0242] Obtaining signal energies of a plurality of sampling points of the input signal when the to-be-tested LED lamp is in the off state;

[0243] Calculating an average value of the signal energies of the plurality of sampling points to obtain the static energy value.

[0244] In some preferred embodiments, the obtaining of the signal energy of the input signal at the moment when the to-be-tested LED lamp is turned on to obtain the dynamic energy value comprises:

[0245] Determining a target sampling point corresponding to the moment when the to-be-tested LED lamp is turned on;

[0246] Obtaining a signal energy of the input signal at the target sampling point to obtain the dynamic energy value.

[0247] In some preferred embodiments, the determining whether the to-be-tested LED lamp is a normal LED lamp based on the input signal comprises:

[0248] Performing feature extraction on the input signal to obtain an input feature;

[0249] Determining whether the to-be-tested LED lamp is a normal LED lamp according to the input feature through a pre-trained machine learning model.

[0250] The storage medium is a physical, non-transient storage medium, for example, can be a U disk, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk, and various physical storage media that can store program codes. The computer-readable storage medium can be non-volatile or volatile.

[0251] Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been described in the above description in a general manner. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0252] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of each unit is only a logical function division, and actual implementation can have another division manner. For example, a plurality of units or components can be combined or integrated into another system, or some features can be omitted or not executed.

[0253] The steps in the method embodiments of the present application can be adjusted, combined and deleted in sequence according to actual needs. The units in the device embodiments of the present application can be combined, divided and deleted according to actual needs. In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit.

[0254] The integrated unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art, or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a plurality of instructions for causing a computer device (which can be a personal computer, a terminal or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application.

[0255] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0256] It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application cover modifications and variations of this application provided they come within the scope of the appended claims and their equivalents.

[0257] The above descriptions are only the specific embodiments of the application, but the protection scope of the application is not limited to this. Any skilled person in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the application, and these modifications or replacements should be covered in the protection scope of the application. Therefore, the protection scope of the application should be subject to the protection scope of the claims.

Claims

1. A method for detecting LED bad pixels based on laser reflection, characterized in that: include: Control the laser light source to irradiate the laser towards the LED lamp to be tested; Collecting the reflected light signal reflected by the LED lamp to be tested; Preprocessing the reflected light signal to obtain an input signal; It is determined whether the LED lamp to be tested is a normal LED lamp based on the input signal.

2. The LED bad pixel detection method based on laser reflection according to claim 1, characterized in that: The controlling the laser light source to irradiate the laser light to the LED lamp to be tested comprises: Get the preset PWM signal; The laser light source is controlled based on the PWM signal to irradiate laser light toward the LED lamp to be tested.

3. The LED bad pixel detection method based on laser reflection according to claim 1, characterized in that: The preprocessing of the reflected light signal to obtain an input signal includes: The reflected light signal is subjected to noise reduction processing to obtain the input signal, wherein the noise reduction processing includes wavelet transform denoising.

4. The method for detecting LED bad pixels based on laser reflection according to claim 1, wherein: The determining whether the LED lamp to be tested is a normal LED lamp based on the input signal includes: Acquire the signal energy of the input signal when the LED lamp to be tested is in the off state to obtain a static energy value; Obtain the signal energy of the input signal at the moment when the LED lamp to be tested is turned on to obtain a dynamic energy value; Obtaining a ratio of the dynamic energy value to the static energy value; Determining whether the ratio is greater than a preset ratio threshold; If the ratio is greater than a preset ratio threshold, it is determined that the LED lamp to be tested is a normal LED lamp.

5. The LED bad pixel detection method based on laser reflection according to claim 4, characterized in that: The step of obtaining the signal energy of the input signal when the LED lamp to be tested is in the off state to obtain a static energy value includes: Acquire signal energy of a plurality of sampling points of the input signal when the LED lamp to be tested is in an off state; An average value of the signal energies of the plurality of sampling points is calculated to obtain the static energy value.

6. The method for detecting LED bad pixels based on laser reflection according to claim 4, characterized in that: The step of obtaining the signal energy of the input signal at the moment when the LED lamp to be tested is turned on to obtain a dynamic energy value includes: Determine the target sampling point corresponding to the moment when the LED lamp to be tested is turned on; The signal energy of the input signal at the target sampling point is acquired to obtain the dynamic energy value.

7. The LED bad pixel detection method based on laser reflection according to claim 1, characterized in that: The determining whether the LED lamp to be tested is a normal LED lamp based on the input signal includes: Performing feature extraction on the input signal to obtain input features; A pre-trained machine learning model is used to determine whether the LED lamp to be tested is a normal LED lamp based on the input features.

8. A device for detecting LED bad pixels based on laser reflection, characterized in that: The method comprises means for performing the method according to any one of claims 1 to 7.

9. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that The storage medium stores a computer program, and when the computer program is executed by a processor, the computer program can implement the method according to any one of claims 1 to 7.

Images