Acousto-optic fusion detection method for bubble defects in display screen panel film layer bonding

Abstract

The present application relates to the technical field of display screen panel detection, and discloses a sound-light fusion detection method for bubble defects of a display screen panel film layer, comprising the following steps: first, using an acoustic phased array to emit an initial global sound wave to a display screen panel to be measured, and using a stroboscopic optical system that is in phase with the initial global sound wave, continuously collecting a full-field interference image sequence containing surface sound wave deformation and optical interference fringes of the display screen panel to be measured; then, converting the full-field interference image sequence into a one-dimensional vector and arranging it in time sequence. By constructing a time-space coherence data matrix from the collected full-field interference image sequence, and introducing a dynamic mode decomposition operation to solve a linear evolution operator, false point misjudgment phenomena occurring in conventional optical detection are avoided, and by extracting specific feature vectors for non-light-transmitting blind areas from high-frequency dynamic space modes, the problem of bubble detection leakage in blind areas caused by the inability of conventional optical means to penetrate the obstruction is solved.

Description

Technical Field

[0001] This invention relates to the field of display panel inspection technology, specifically to an acoustic-optical fusion detection method for bubble defects in the film bonding of display panels. Background Technology

[0002] With the rapid development of new display technologies, high-end display panels such as OLED, LCD, and foldable screens are increasingly widely used in smartphones, automotive displays, and wearable devices. In the manufacturing process of display panels, a multi-layer film bonding operation is usually required. For example, cover glass, polarizer, and touch layer are laminated and bonded using OCA optically transparent adhesive. However, due to the limitations of the production environment, material properties, and process fluctuations, tiny air bubbles are easily left at the film bonding interface. These bonding bubbles not only affect the appearance quality and optical display effect of the panel, but may also cause the film layer to detach or even the panel to fail due to stress concentration during subsequent use. Therefore, high-speed and high-precision online detection of bonding bubble defects is a key link to ensure the yield rate of panel products leaving the factory.

[0003] Currently, the industry mainly relies on optical machine vision (AOI) technology to detect bubbles in the bonding of display panels. This technology uses light sources of different angles or wavelengths to illuminate the panel surface and uses an industrial camera to capture the abnormal features of light refraction or total internal reflection caused by bubbles for image judgment. Although optical inspection has the advantages of non-contact and fast full-field imaging, with the increasing complexity of panel structures, the superposition of multiple layers of transparent materials often produces severe multiple reflections and light interference phenomena under light illumination. These static optical ghosts and interference fringes in the acquired two-dimensional images are extremely similar to micron-sized bubbles, making it difficult for conventional optical image processing algorithms to completely separate the two, thus causing false spot misjudgments. At the same time, modern display panels usually have black ink areas printed on the edges to cover the internal wiring. The optical line of sight cannot penetrate these opaque materials, making the bonding bubbles hidden under the black edge a complete blind spot for detection, which is very easy to miss. Therefore, this invention designs an acousto-optical fusion detection method for bubble defects in the bonding of film layers of display panels to address the above-mentioned problems. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an acoustic-optical fusion detection method for bubble defects in the film bonding of display panels. This method solves the problems of optical interference misjudgment caused by the superposition of multiple transparent films and missed detection of bubble defects caused by the opaque black edge blind zone, which are difficult to overcome in existing detection technologies.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel, comprising the following steps:

[0006] S1. First, an initial global acoustic wave is emitted from the display panel under test using an acoustic phased array, and then a full-field interference image sequence containing the acoustic waveform changes and optical interference fringes on the surface of the display panel under test is continuously acquired using a stroboscope optical system that is phase-locked with the initial global acoustic wave.

[0007] S2. Subsequently, the full-field interferometric image sequence is converted into a one-dimensional vector and arranged in the order of acquisition time to construct a spatiotemporal coherent data matrix with time step delay.

[0008] S3. Perform dynamic mode decomposition operation on the spatiotemporal coherent data matrix to solve the linear evolution operator between the spatiotemporal coherent data matrices, so as to decompose the low-frequency mode characterizing the optical interference fringes and the high-frequency dynamic spatial mode characterizing the suspected film bonding bubble defect. Remove the low-frequency mode and output the initial detection information of bubble defects in the transparent area of ​​the display panel under test according to the retained high-frequency dynamic spatial mode.

[0009] S4. Extract specific feature vectors for the opaque blind zone within the display panel under test from the high-frequency dynamic spatial modes, calculate the complex conjugate matrix of the specific feature vectors in the mathematical dimension, map the complex conjugate matrix to the control parameters of the acoustic phased array, and emit reverse customized sound waves to the display panel under test through the acoustic phased array, so that the acoustic energy of the reverse customized sound waves is targeted and focused on the position of the weak bubble under the opaque blind zone.

[0010] S5. At the instant when the acoustic energy of the reversed custom sound wave is focused on the position of the weak bubble, the stroboscope optical system is used again to perform full-field synchronous shooting to obtain the target bubble defect image. The blind zone bubble information extracted from the target bubble defect image is fused with the initial bubble defect detection information, and finally the detection result of the complete film bonding bubble defect of the display panel under test is output.

[0011] Preferably, step S1 specifically includes:

[0012] A multi-channel air-coupled ultrasonic phased array arranged at the edge of the display panel under test is used to excite a broadband Lamb wave into the interior of the panel as the initial global acoustic wave.

[0013] The pulse trigger frequency of the high-frequency stroboscopic structured light illumination system covering the entire field of the panel is phase-locked with the excitation frequency of the multi-channel air-coupled ultrasonic phased array to capture the dynamic phase changes on the surface of the display panel under test in a non-contact manner and generate the full-field interference image sequence.

[0014] Preferably, step S2 specifically includes:

[0015] The acquired single-frame two-dimensional full-field interferometric image is flattened into a one-dimensional column vector in a fixed order;

[0016] According to the time acquisition order, the one-dimensional column vectors corresponding to adjacent frames are combined to construct a basic spatiotemporal matrix and a delayed spatiotemporal matrix that is delayed by one time step compared to the basic spatiotemporal matrix. The basic spatiotemporal matrix and the delayed spatiotemporal matrix are used together as the spatiotemporal coherent data matrix.

[0017] Preferably, step S3 specifically includes:

[0018] Singular value decomposition is performed on the basic spatiotemporal matrix to achieve dimensionality reduction, and the dimensionality-reduced linear evolution operator projected onto the low-dimensional subspace is calculated in combination with the delayed spatiotemporal matrix.

[0019] The eigenvalue decomposition of the dimensionality reduction linear evolution operator is performed to obtain eigenvalues ​​and eigenvectors. The eigenvectors are then restored to the original high-dimensional space to obtain the spatial mode matrix, thereby separating different modes.

[0020] Preferably, the step of discarding the low-frequency modes and outputting initial detection information of bubble defects in the transparent area of ​​the display panel under test based on the retained high-frequency dynamic spatial modes specifically includes:

[0021] The position of the eigenvalue on the complex plane is determined, and the mode corresponding to the frequency located on the unit circle of the complex plane or close to zero frequency is determined as the low-frequency mode characterizing the optical interference fringe, and is set to zero and removed from the spatial mode matrix;

[0022] The spatial distribution coordinates corresponding to the remaining non-zero high-frequency dynamic spatial modes are analyzed and directly output as the initial detection information of the bubble defects in the transparent area.

[0023] Preferably, the opaque blind zone is the black ink area at the edge of the display panel under test; extracting specific feature vectors for the opaque blind zone within the display panel under test from high-frequency dynamic spatial modes specifically includes:

[0024] The vector representing the spatial phase and amplitude distribution of the initial global acoustic wave propagating to the panel surface in the opaque blind zone due to scattering by weak bubbles is extracted from the high-frequency dynamic spatial mode and used as the specific feature vector.

[0025] Preferably, calculating the complex conjugate matrix of the specific eigenvector in the mathematical dimension specifically includes:

[0026] The extracted specific feature vectors are subjected to complex conjugate mathematical operations to generate the complex conjugate matrix, so as to reverse the evolution and mapping of the spatial phase containing the scattering information of the bubble defect in the film layer on the mathematical time axis, and generate the basic data for driving the hardware to produce sound.

[0027] Preferably, the step of mapping the complex conjugate matrix to the control parameters of the acoustic phased array specifically includes:

[0028] Through digital-to-analog conversion, the inverted spatial phase and amplitude information in the complex conjugate matrix is ​​parsed and mapped to the delay time parameters and excitation voltage amplitude parameters of each independent transmission channel in the acoustic phased array, which are then used as the control parameters.

[0029] Preferably, step S5 specifically includes:

[0030] Each independent transmission channel transmits the reversed customized acoustic wave according to the control parameters. Based on the acoustic time reversal law, the reversed customized acoustic wave of each channel propagates backward along the initial scattering path in the medium of the display panel under test.

[0031] Multiple reverse-propagating sound waves coherently superimpose at the location of the weak bubble, forming a local physical acoustic energy standing wave resonance, thereby amplifying the surface vibration ripples at the location of the weak bubble.

[0032] Preferably, the step of fusing the blind zone bubble information extracted from the targeted bubble defect image with the initial bubble defect detection information to output a complete film bonding bubble defect detection result for the display panel under test specifically includes:

[0033] The spatial coordinates and contours of the blind zone bubble information extracted from the targeted bubble defect image are aligned and merged with the spatial coordinates and contours of the initial bubble defect detection information in the global spatial coordinate system.

[0034] A complete defect map is generated and output, taking into account both transparent and opaque areas while filtering out optical interference, as the detection result of the bubble defect in the film bonding.

[0035] This invention provides an acoustic-optical fusion detection method for bubble defects in the film bonding of display panels. It has the following beneficial effects:

[0036] 1. This invention constructs the acquired full-field interference image sequence into a spatiotemporal coherent data matrix with a time step delay, and introduces dynamic mode decomposition operation across domains to solve the linear evolution operator. This enables orthogonal decoupling of static or extremely low-frequency optical interference fringe features with high-frequency dynamic acoustic distortion features from a mathematical perspective, effectively avoiding the false spot misjudgment phenomenon that is very easy to occur in conventional optical detection.

[0037] 2. This invention solves the problem of conventional optical methods being unable to penetrate obstructions and conventional signal extraction algorithms failing to detect bubbles in blind zones due to strong damping attenuation by specifically extracting specific feature vectors for opaque blind zones from high-frequency dynamic spatial modes and calculating their complex conjugate matrix in a mathematical dimension.

[0038] 3. This invention employs a multi-channel air-coupled ultrasonic phased array to excite an initial global acoustic wave onto the panel, and combines this with a stroboscope optical system that is strictly phase-locked with the acoustic wave to continuously acquire full-field interference images containing surface acoustic wave variations. This ingeniously integrates the penetrability of acoustic waves in probing deep media with the ability of optical full-section arrays to rapidly image. It eliminates the risk of physical contamination to high-cleanliness display panels caused by the liquid coupling agent required for traditional high-frequency ultrasonic testing, and also avoids the inefficiency limitations of point-by-point mechanical scanning in traditional acoustic flaw detection. This greatly matches the high-speed requirements of large-scale assembly line production of display panels.

[0039] 4. This invention designs an acoustic physical energy closed-loop control mechanism that combines "initial global low-power detection with secondary local targeted high-energy focusing and re-enhancing". In the first stage, only extremely low global emission power is required for the algorithm to capture the weak scattering phase of the blind zone. In the second stage, through feedback mapping of control parameters, the high-intensity acoustic standing wave resonance is precisely confined to the in-situ local space of the micro bubble defect. This effectively protects the fragile flexible organic light-emitting layer and precision wiring inside the display panel from damage caused by large-area strong sound field vibration, ensuring the absolute non-destructive safety of the detection.

[0040] 5. This invention utilizes an optical system to directly read the dynamic spatial phase of the panel surface modulated by ultrasound, and converts the optical image sequence into a spatiotemporal coherent data matrix for decomposition. This allows for the reverse calculation of conjugate control parameters used to directly drive the physical hardware to produce sound. This breaks through the conventional technical barrier of traditional multimodal detection schemes, where "optics and acoustics are detected independently, and images and waveform data are forcibly stitched together in the background." It achieves true acoustic-optical closed-loop collaborative operation in both the physical signal source and mathematical algorithm evolution dimensions, avoiding the technical bottleneck of spatial coordinate registration of heterogeneous data at the micrometer level, and significantly improving the reliability and accuracy of global defect map fusion. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the method architecture of the present invention;

[0042] Figure 2 This is one of the schematic diagrams of the method flow of the present invention;

[0043] Figure 3 This is a second schematic diagram of the method flow of the present invention;

[0044] Figure 4 This is the third schematic diagram of the method flow of the present invention;

[0045] Figure 5 This is the fourth schematic diagram of the method flow of the present invention;

[0046] Figure 6 This is the fifth schematic diagram of the method flow of the present invention. Detailed Implementation

[0047] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Please see the appendix Figure 1 -Appendix Figure 6 This invention provides an acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel, comprising the following steps:

[0049] S1. First, an initial global acoustic wave is emitted from the display panel under test using an acoustic phased array. Then, a stroboscopic optical system phase-locked with the initial global acoustic wave is used to continuously acquire a full-field interference image sequence containing the acoustic waveform changes and optical interference fringes on the surface of the display panel under test. Specifically, step S1 includes: using a multi-channel air-coupled ultrasonic phased array arranged at the edge of the display panel under test to excite a broadband Lamb wave into the panel as the initial global acoustic wave; and phase-locking the pulse trigger frequency of the high-frequency stroboscopic structured light illumination system covering the entire field of the panel with the excitation frequency of the multi-channel air-coupled ultrasonic phased array to capture the dynamic phase changes on the surface of the display panel under test in a non-contact manner and generate a full-field interference image sequence.

[0050] Specifically, the core of this step lies in achieving full-field non-contact acousto-optic phase modulation at the physical level. In practice, the display panel under test (such as an OLED or LCD panel with pre-bonded OCA adhesive, polarizer, and other films) is placed on the testing platform. A multi-channel air-coupled ultrasonic phased array arranged at the edge of the display panel excites a broadband Lamb wave into the panel. Using a low-power emission wave (wave) to avoid damaging the delicate internal structure of the panel, this initial global acoustic wave propagates through the panel's multiple layers. When it encounters tiny air bubbles (containing air and with extremely low acoustic impedance) at the bonding interface, strong local total internal reflection and scattering occur, resulting in micron-sized dynamic standing wave ripples (i.e., surface acoustic wave variations) on the outermost surface of the display panel under test. Simultaneously, a high-frequency stroboscopic structured light illumination system covering the entire panel is activated. To stably capture the dynamic micro-ripples of high-frequency vibrations, the pulse trigger frequency of the stroboscopic light source is strictly phase-locked with the excitation frequency of the multi-channel air-coupled ultrasonic phased array. A high-speed industrial camera continuously captures images within one or more acoustic wave propagation cycles to obtain a full-field interference image sequence. The acquired image sequence contains not only the surface dynamic phase changes caused by internal air bubbles (useful high-frequency signals) but also interference fringes (useless low-frequency / static noise) generated by the superposition of multiple transparent film layers on the panel.

[0051] S2. Subsequently, the full-field interferometric image sequence is converted into a one-dimensional vector and arranged in the order of acquisition time to construct a spatiotemporal coherent data matrix with a time step delay. Specifically, step S2 includes: flattening the acquired single-frame two-dimensional full-field interferometric image into a one-dimensional column vector in a fixed order; combining the one-dimensional column vectors corresponding to adjacent frames according to the time acquisition order to construct a basic spatiotemporal matrix and a delayed spatiotemporal matrix that is delayed by one time step compared to the basic spatiotemporal matrix; and using the basic spatiotemporal matrix and the delayed spatiotemporal matrix together as the spatiotemporal coherent data matrix.

[0052] Specifically, this step aims to transform the acquired physical image stream into a spatiotemporal data structure suitable for mathematical dimensionality reduction and orthogonal decomposition, setting the spatial resolution of a single-frame full-field interferometric image acquired by a high-speed industrial camera to be [value missing]. First, the continuously collected data The two-dimensional interferometric images are flattened one by one in a fixed row and column order, and transformed into a dimension of one-dimensional column vector ,in ,and Subsequently, image column vectors from adjacent time nodes are extracted and combined according to the time acquisition sequence to construct a basic spatiotemporal matrix. And a delayed spatiotemporal matrix that is delayed by one time step (i.e., one frame interval) compared to the basic spatiotemporal matrix. Its mathematical expression is:

[0053] Basic spacetime matrix: ;

[0054] Delayed spacetime matrix: ;

[0055] These two matrices contain the spatial full-field information of the system in the column dimension and record the system's evolution state over time in the row dimension. Together, they serve as spatiotemporal coherent data matrices input into subsequent algorithms.

[0056] S3. Perform dynamic mode decomposition on the spatiotemporal coherence data matrix to solve for the linear evolution operator between the spatiotemporal coherence data matrices, thereby decomposing the low-frequency modes characterizing optical interference fringes and the high-frequency dynamic spatial modes characterizing suspected film-bonded bubble defects. Discard the low-frequency modes and output the initial detection information of bubble defects in the transparent area of ​​the display panel under test based on the retained high-frequency dynamic spatial modes. Specifically, step S3 includes: performing singular value decomposition on the basic spatiotemporal matrix to achieve dimensionality reduction, and calculating the dimensionality-reduced linear evolution operator projected onto the low-dimensional subspace using the delayed spatiotemporal matrix; performing eigenvalue decomposition on the dimensionality-reduced linear evolution operator to obtain... The process involves identifying eigenvalues ​​and eigenvectors, then restoring the eigenvectors to the original high-dimensional space to obtain a spatial mode matrix to separate different modes. Low-frequency modes are eliminated, and the initial detection information for bubble defects in the transparent area of ​​the display panel under test is output based on the retained high-frequency dynamic spatial modes. Specifically, this includes: determining the position of the eigenvalues ​​on the complex plane; identifying modes located on the unit circle of the complex plane or close to zero frequency as low-frequency modes that characterize optical interference fringes, and eliminating them from the spatial mode matrix; and analyzing the spatial distribution coordinates corresponding to the remaining non-zero high-frequency dynamic spatial modes, directly outputting the initial detection information for bubble defects in the transparent area.

[0057] Specifically, this step utilizes Dynamic Mode Decomposition (DMD) to mathematically decouple the image across domains, assuming that the system state follows linear evolution, i.e., that a linear evolution operator exists. satisfy ;

[0058] Firstly, due to the extremely large dimension of the original matrix, in order to improve computational feasibility, the basic spatiotemporal matrix is ​​modified. Singular Value Decomposition (SVD) is performed to achieve dimensionality reduction. The formula is as follows: ,in and It is an orthogonal matrix. is a singular value diagonal matrix, and * denotes the complex conjugate transpose;

[0059] Then, combined with the delayed spatiotemporal matrix Calculate the projection onto the object. Dimensionality reduction linear evolution operator in a low-dimensional subspace constructed by truncation The calculation formula is: ;

[0060] Then, the dimension reduction linear evolution operator is applied. Perform eigenvalue decomposition: The diagonal matrix It includes the complex eigenvalues ​​of the system, matrix This includes the corresponding feature vectors, which are then restored from low-dimensional feature vectors to the original high-dimensional space to obtain the spatial mode matrix. After separating the modes, the eigenvalue diagonal matrix is ​​determined. Information is filtered based on the positions of each eigenvalue on the complex plane. The static interference fringes generated by the multilayer transparent film are time-invariant, corresponding to eigenvalues ​​on the unit circle of the complex plane with phase angles close to 0 (i.e., zero-frequency or extremely low-frequency modes). These low-frequency modes characterizing the optical interference fringes are then analyzed in the spatial mode matrix. The high-frequency dynamic spatial modes corresponding to the remaining non-zero eigenvalues ​​are forcibly set to zero and removed. These modes characterize the suspected bubble defect signals generated by acoustic wave modulation. The spatial distribution coordinates of the retained high-frequency dynamic spatial modes can be analyzed to directly output the initial detection information of bubble defects in the transparent area of ​​the display panel under test, which can completely eliminate interference misjudgment.

[0061] S4. Extract specific feature vectors for the opaque blind zone within the display panel under test from the high-frequency dynamic spatial modes, and calculate the complex conjugate matrix of the specific feature vectors mathematically. Map the complex conjugate matrix to the control parameters of the acoustic phased array. Emit a reversed customized acoustic wave to the display panel under test through the acoustic phased array, so that the acoustic energy of the reversed customized acoustic wave is targeted and focused on the position of the weak bubble under the opaque blind zone. The opaque blind zone is the black ink area at the edge of the display panel under test. Extracting specific feature vectors for the opaque blind zone within the display panel under test from the high-frequency dynamic spatial modes specifically includes: extracting the feature vectors representing the propagation of the initial global acoustic wave in the opaque blind zone due to scattering by the weak bubble. The vector of spatial phase and amplitude distribution formed on the plate surface is used as a specific feature vector. The complex conjugate matrix of the specific feature vector is calculated in the mathematical dimension. Specifically, this includes: performing complex conjugate mathematical operations on the extracted specific feature vector to generate a complex conjugate matrix, so as to reverse the evolution and mapping of the spatial phase containing the scattering information of the bubble defect in the film bonding on the mathematical time axis, generating the basic data for driving the hardware to produce sound, and mapping the complex conjugate matrix into the control parameters of the acoustic phased array. Specifically, this includes: through digital-to-analog conversion, parsing and mapping the inverted spatial phase and amplitude information in the complex conjugate matrix into the delay time parameters and excitation voltage amplitude parameters of each independent transmission channel in the acoustic phased array, as control parameters.

[0062] Specifically, for the area around the panel with black ink (i.e., the opaque blind zone), due to the high damping and opaque properties of the material, although the sound waves generated by the bubbles at the bottom can cause surface deformation in the blind zone, this deformation is extremely weak in the high-frequency dynamic spatial mode extracted in step S3, making it difficult to directly characterize using conventional thresholds. At this point, the detection perspective shifts from macroscopic identification to microscopic extraction. Based on the known physical dimensions of the display panel, the data area corresponding to the opaque blind zone position in the high-frequency dynamic spatial mode is defined. From this, a vector representing the spatial phase and amplitude distribution formed by the initial global sound wave propagating to the panel surface in the blind zone due to scattering by the weak bubbles is extracted as a specific feature vector. Then, on a mathematical dimension, a complex conjugate operation is performed on this specific eigenvector: The physical essence of this mathematical operation is that the spatial phase matrix of the weak scattered sound wave when it reaches the surface is reversed in time, providing a core basic mathematical model for the subsequent reconstruction of the reverse propagation sound field.

[0063] The pure mathematical calculations are then fed back to the physical actuators, forming a closed hardware-software loop. The generated complex conjugate matrix is ​​then converted into an analog-to-digital (D / A) conversion module. The spatial phase and amplitude information after inversion is analyzed and mapped to the control parameters of each independent transmission channel in the acoustic phased array (including the trigger delay time parameter and excitation voltage amplitude parameter of each piezoelectric element). Based on the above control parameters, the acoustic phased array emits inverted customized acoustic waves to the display panel under test. According to the acoustic time reversal law, the wave equation in the physical medium is symmetrical with respect to time. Therefore, the inverted customized acoustic waves emitted by each channel will strictly propagate in reverse along the path of initial scattering in the complex layered medium of the display panel under test. Finally, the multiple reverse propagating acoustic waves will spatially coherently superimpose at the original position of the weak bubble below the opaque blind zone, automatically forming a local physical acoustic energy standing wave resonance. This targeted focusing is concentrated in a very small area, greatly amplifying the surface vibration ripples at the position of the weak bubble, while the rest of the healthy area of ​​the panel is not affected by the strong acoustic energy.

[0064] S5. At the instant the acoustic energy of the reversed custom acoustic wave is focused on the location of the weak bubble, a full-field synchronous imaging is performed again using a stroboscope optical system to obtain a target bubble defect image. The blind zone bubble information extracted from the target bubble defect image is fused with the initial bubble defect detection information. Finally, the detection result of the complete film bonding bubble defect of the display panel under test is output. Specifically, step S5 includes: each independent transmission channel emits a reversed custom acoustic wave according to the control parameters. Based on the acoustic time reversal law, the reversed custom acoustic waves of each channel propagate backward along the initial scattering path in the medium of the display panel under test; the multi-path backward propagating acoustic waves at the location of the weak bubble... Spatial coherent superposition occurs at the location, forming local physical acoustic energy standing wave resonance, thereby amplifying the surface vibration ripples at the location of the weak bubble. The blind zone bubble information extracted from the targeted bubble defect image is fused with the initial bubble defect detection information to output a complete film bonding bubble defect detection result for the display panel under test. Specifically, this includes: aligning and merging the spatial coordinates and contours of the blind zone bubble information extracted from the targeted bubble defect image with the spatial coordinates and contours of the initial bubble defect detection information in the global spatial coordinate system; generating and outputting a complete defect map that takes into account both transparent and opaque blind zones and filters out optical interference interference, as the film bonding bubble defect detection result;

[0065] Specifically, at the peak moment of physical energy targeting and focusing, the system control terminal triggers the stroboscopic optical system and industrial camera to perform a second full-field synchronous shooting to acquire the targeted bubble defect image. In this image, the bubble, which was originally hidden under the opaque blind zone and had an extremely weak initial signal, exhibits an extremely high signal-to-noise ratio due to the strong physical drive of the reversed customized acoustic wave. Finally, the data processing system extracts the high signal-to-noise ratio blind zone bubble coordinates and contours from the targeted bubble defect image and aligns and merges them with the initial bubble defect detection information (transparent area bubble data) output in step S3 in the global spatial coordinate system. Finally, a complete defect map covering all areas of the display panel under test (including transparent areas and opaque blind zones) and unaffected by multilayer film optical interference is generated and output as the final film bonding bubble defect detection result.

[0066] In summary, this invention provides an acousto-optic fusion detection method for bubble defects in the film bonding of display panels. By constructing the acquired full-field interference image sequence into a spatiotemporal coherent data matrix with a time step delay, and introducing dynamic mode decomposition operations across domains to solve linear evolution operators, it can orthogonally decouple the static or extremely low-frequency optical interference fringe features from the high-frequency dynamic acoustic distortion features from a mathematical perspective. By directly zeroing out the low-frequency modes representing optical interference fringes in the spatial mode matrix, the interference of optical ghosting generated by the superposition of multiple transparent films on the display panel on defect feature identification is fundamentally eliminated, effectively avoiding the interference of conventional optical detection. The problem of false spot misjudgment is very common. By extracting specific feature vectors for opaque blind zones from high-frequency dynamic spatial modes and calculating their complex conjugate matrix in the mathematical dimension to map them as control parameters of the acoustic phased array, the array is driven to emit reverse customized sound waves. The physical law of acoustic time reversal is used to make the emitted sound waves automatically propagate in reverse along the scattering path of the weak bubbles and undergo spatial coherent superposition. At the physical medium level, targeted resonant amplification of the energy of weak defects below opaque blind zones such as black edges is achieved. This solves the problem that conventional optical methods cannot penetrate obstructions and conventional signal extraction algorithms face the problem of missed detection of bubbles in blind zones caused by strong damping attenuation.

[0067] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel, characterized in that, Includes the following steps: S1. First, an initial global acoustic wave is emitted from the display panel under test using an acoustic phased array, and then a full-field interference image sequence containing the acoustic waveform changes and optical interference fringes on the surface of the display panel under test is continuously acquired using a stroboscope optical system that is phase-locked with the initial global acoustic wave. S2. Subsequently, the full-field interferometric image sequence is converted into a one-dimensional vector and arranged in the order of acquisition time to construct a spatiotemporal coherent data matrix with time step delay. S3. Perform dynamic mode decomposition operation on the spatiotemporal coherent data matrix to solve the linear evolution operator between the spatiotemporal coherent data matrices, so as to decompose the low-frequency mode characterizing the optical interference fringes and the high-frequency dynamic spatial mode characterizing the suspected film bonding bubble defect. Remove the low-frequency mode and output the initial detection information of bubble defects in the transparent area of ​​the display panel under test according to the retained high-frequency dynamic spatial mode. S4. Extract specific feature vectors for the opaque blind zone within the display panel under test from the high-frequency dynamic spatial modes, calculate the complex conjugate matrix of the specific feature vectors in the mathematical dimension, map the complex conjugate matrix to the control parameters of the acoustic phased array, and emit reverse customized sound waves to the display panel under test through the acoustic phased array, so that the acoustic energy of the reverse customized sound waves is targeted and focused on the position of the weak bubble under the opaque blind zone. S5. At the instant when the acoustic energy of the reversed custom sound wave is focused on the position of the weak bubble, the stroboscope optical system is used again to perform full-field synchronous shooting to obtain the target bubble defect image. The blind zone bubble information extracted from the target bubble defect image is fused with the initial bubble defect detection information, and finally the detection result of the complete film bonding bubble defect of the display panel under test is output.

2. The acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel according to claim 1, characterized in that, Step S1 specifically includes: A multi-channel air-coupled ultrasonic phased array arranged at the edge of the display panel under test is used to excite a broadband Lamb wave into the interior of the panel as the initial global acoustic wave. The pulse trigger frequency of the high-frequency stroboscopic structured light illumination system covering the entire field of the panel is phase-locked with the excitation frequency of the multi-channel air-coupled ultrasonic phased array to capture the dynamic phase changes on the surface of the display panel under test in a non-contact manner and generate the full-field interference image sequence.

3. The acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel according to claim 1, characterized in that, Step S2 specifically includes: The acquired single-frame two-dimensional full-field interferometric image is flattened into a one-dimensional column vector in a fixed order; According to the time acquisition order, the one-dimensional column vectors corresponding to adjacent frames are combined to construct a basic spatiotemporal matrix and a delayed spatiotemporal matrix that is delayed by one time step compared to the basic spatiotemporal matrix. The basic spatiotemporal matrix and the delayed spatiotemporal matrix are used together as the spatiotemporal coherent data matrix.

4. The acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel according to claim 1, characterized in that, Step S3 specifically includes: Singular value decomposition is performed on the basic spatiotemporal matrix to achieve dimensionality reduction, and the dimensionality-reduced linear evolution operator projected onto the low-dimensional subspace is calculated in combination with the delayed spatiotemporal matrix. The eigenvalue decomposition of the dimensionality reduction linear evolution operator is performed to obtain eigenvalues ​​and eigenvectors. The eigenvectors are then restored to the original high-dimensional space to obtain the spatial mode matrix, thereby separating different modes.

5. The acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel according to claim 1, characterized in that, The process of removing the low-frequency modes and outputting initial detection information of bubble defects in the transparent area of ​​the display panel under test based on the retained high-frequency dynamic spatial modes specifically includes: The position of the eigenvalue on the complex plane is determined, and the mode corresponding to the unit circle of the complex plane or close to zero frequency is determined as the low-frequency mode characterizing the optical interference fringe, and is set to zero and removed from the spatial mode matrix. The spatial distribution coordinates corresponding to the remaining non-zero high-frequency dynamic spatial modes are analyzed and directly output as the initial detection information of the bubble defects in the transparent area.

6. The acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel according to claim 1, characterized in that, The opaque blind zone is the black ink area at the edge of the display panel under test; specific feature vectors for the opaque blind zone within the display panel under test are extracted from the high-frequency dynamic spatial modes, specifically including: The vector representing the spatial phase and amplitude distribution of the initial global acoustic wave propagating to the panel surface in the opaque blind zone due to scattering by weak bubbles is extracted from the high-frequency dynamic spatial mode and used as the specific feature vector.

7. The acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel according to claim 1, characterized in that, The calculation of the complex conjugate matrix of the specific eigenvector in the mathematical dimension specifically includes: The extracted specific feature vectors are subjected to complex conjugate mathematical operations to generate the complex conjugate matrix, so as to perform reverse evolution mapping of the spatial phase containing the scattering information of the bubble defect in the film layer on the mathematical time axis, and generate the basic data for driving the hardware to produce sound.

8. The acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel according to claim 1, characterized in that, The control parameters for mapping the complex conjugate matrix to the acoustic phased array specifically include: Through digital-to-analog conversion, the inverted spatial phase and amplitude information in the complex conjugate matrix is ​​parsed and mapped to the delay time parameters and excitation voltage amplitude parameters of each independent transmission channel in the acoustic phased array, which are then used as the control parameters.

9. The acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel according to claim 1, characterized in that, Step S5 specifically includes: Each independent transmission channel transmits the reversed customized acoustic wave according to the control parameters. Based on the acoustic time reversal law, the reversed customized acoustic wave of each channel propagates backward along the initial scattering path in the medium of the display panel under test. Multiple reverse-propagating sound waves coherently superimpose at the location of the weak bubble, forming a local physical acoustic energy standing wave resonance, thereby amplifying the surface vibration ripples at the location of the weak bubble.

10. The acoustic-optical fusion detection method for bubble defects in the film bonding of a display panel according to claim 1, characterized in that, The process of fusing the blind zone bubble information extracted from the targeted bubble defect image with the initial bubble defect detection information to output a complete film bonding bubble defect detection result for the display panel under test specifically includes: The spatial coordinates and contours of the blind zone bubble information extracted from the targeted bubble defect image are aligned and merged with the spatial coordinates and contours of the initial bubble defect detection information in the global spatial coordinate system. A complete defect map is generated and output, taking into account both transparent and opaque areas while filtering out optical interference, as the detection result of the bubble defect in the film bonding.

Images