Intelligent visual inspection system for high-precision printed circuit board

Abstract

The invention discloses an intelligent visual detection system for a high-precision printed circuit board, and relates to the technical field of precision optical detection.The intelligent visual detection system comprises the steps that based on a printed circuit board substrate with a quantum dot mark layer attached to the surface, confocal scanning is conducted through laser wavelength, and an original image data set is generated; performing multispectral feature analysis on the original image data set, capturing focus offset rules of light rays with different wavelengths, and generating a wavelength-focus mapping relation compensation instruction set; according to the compensation instruction set of the wavelength-focus mapping relation, an imaging light path is adjusted in real time, and a high-fidelity fusion image is generated; and analyzing optical signal feature changes in the high-fidelity fusion image, identifying an abnormal region, extracting microdefect features, and generating a detection report containing the defect features. According to the invention, through uniform dispersion of the quantum dot modified ink in the inert atmosphere and a plasma substrate activation process, molecular-level combination of the solder mask layer and the fluorescent marking material is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of precision optical detection, in particular to an intelligent visual detection system for a high-precision printed circuit board. Background Art

[0002] The current development status of printed circuit board inspection technology shows that with the miniaturization of electronic equipment and the popularization of 5G high-frequency and high-speed applications, line width and line spacing have entered the microscopic scale, promoting the innovation of inspection technology; the current mainstream technologies include: confocal laser scanning microscopy technology uses point scanning and pinhole filtering to achieve sub-surface tomography, multi-spectral tomography technology uses spectroscopic channel acquisition to achieve inner layer structure reconstruction, and intelligent recognition algorithm technology is based on convolutional neural networks for defect feature extraction and classification identification. The three major technical routes jointly build a modern inspection technology system to support the industry's quality control needs.

[0003] Traditional optical methods are limited by the diffraction limit and the signal-to-noise ratio conflict, resulting in blind spots for detecting microcracks and subtle cracks in buried vias. Ultrasonic testing, however, suffers from severe signal attenuation due to material impedance mismatch, preventing critical quality issues from being effectively captured. Wavelength-dependent focal shift caused by broadband light penetrating multi-layered media leads to spatial mismatches in multispectral channels. Existing compensation mechanisms rely on mechanical displacement adjustments, resulting in response speeds that cannot meet the requirements of high-speed production line inspections, creating a technical bottleneck in manufacturing scenarios with stringent real-time requirements. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a high-precision printed circuit board intelligent visual inspection system to solve the problems of blind spots in micro-defect recognition and real-time compensation technology obstruction.

[0006] The present invention provides a high-precision printed circuit board intelligent visual inspection system, comprising: The substrate marking layer coating module builds a uniform fluorescent marking layer on the surface of the printed circuit board through the coating of quantum dot modified ink and light curing process, thus generating a printed circuit board substrate with a quantum dot marking layer attached to the surface; An image dataset generation module generates an original image dataset by performing confocal scanning using a laser wavelength based on a printed circuit board substrate with a quantum dot marker layer attached to the surface; The focus shift compensation module performs multispectral feature analysis on the original image data set, captures the focus shift patterns of light of different wavelengths, and generates a wavelength-focus mapping relationship compensation instruction set; The imaging optical path adjustment module adjusts the imaging optical path in real time based on the compensation instruction set of the wavelength-focus mapping relationship to generate a high-fidelity fused image; The defect report generation module analyzes the changes in optical signal characteristics in high-fidelity fusion images, identifies abnormal areas and extracts microscopic defect features, and generates an inspection report containing defect features.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: As a preferred solution of the high-precision printed circuit board intelligent visual inspection system of the present invention, wherein: the quantum dot modified ink coating and light curing process, the steps are as follows: The quantum dot material is mixed with the light-curing solder resist ink in an inert atmosphere, and a modified ink with uniform dispersion of quantum dots is obtained through mechanical stirring and cyclic grinding processes; The printed circuit board substrate is subjected to plasma surface activation treatment to obtain a surface activated substrate.

[0008] As a preferred solution of the high-precision printed circuit board intelligent visual inspection system of the present invention, wherein: the steps of constructing a uniform fluorescent marking layer on the surface of the printed circuit board to generate a printed circuit board substrate with a quantum dot marking layer attached to the surface are as follows: Based on the surface activated substrate, the modified ink with uniform dispersion of quantum dots is formed into a wet film coating with uniform thickness through a slot coating process to generate a semi-cured substrate; Based on the application of staged light irradiation to the semi-cured substrate, the modified ink photo-crosslinking reaction is triggered to form a fully cured substrate, and the quantum dots of the fully cured substrate are simultaneously monitored to generate a printed circuit board substrate with a quantum dot marking layer attached to the surface.

[0009] As a preferred solution of the high-precision printed circuit board intelligent visual inspection system of the present invention, wherein: the printed circuit board substrate based on the surface-attached quantum dot marking layer, the steps are as follows: Based on a printed circuit board substrate with a quantum dot marking layer attached to the surface, the quantum dot marking substrate is fixed by vacuum adsorption, and a motion trajectory solution is generated in combination with machine vision positioning; Based on the motion trajectory scheme, the laser source is controlled to emit an excitation beam that matches the intrinsic response of the quantum dots, and point-by-point scanning is performed to generate a fluorescence signal stream.

[0010] As a preferred solution of the intelligent visual inspection system for high-precision printed circuit boards of the present invention, wherein: the laser wavelength is used for confocal scanning to generate the original image data set, the steps are as follows: The wavelength signal of the fluorescence signal stream is separated by spectral filtering, converted into digital coding information in real time, and a digital matrix of associated spatial positions is generated; The digital matrix of associated spatial positions is pixel-mapped and reassembled in the trajectory spatial order, and the original image dataset is generated through data integrity verification.

[0011] As a preferred solution of the high-precision printed circuit board intelligent visual inspection system described in the present invention, the multispectral feature analysis based on the original image data set refers to using a multispectral separation algorithm to extract the fluorescence wavelength channel and the excitation wavelength residual channel, and performing spatial frequency domain analysis to identify the feature point coordinate set.

[0012] As a preferred solution of the intelligent visual inspection system for high-precision printed circuit boards of the present invention, the steps of capturing the focus shift pattern of light of different wavelengths and generating a wavelength-focus mapping relationship compensation instruction set are as follows: Calculate the spatial position offset vector of the same feature point in the feature point coordinate set under the fluorescence wavelength channel and the excitation wavelength channel to generate a wavelength-focus offset characteristic curve; According to the wavelength-focus shift characteristic curve, the focus compensation required for each wavelength is calculated, and the compensation sequence is optimized using the gradient descent method to generate a compensation instruction set of the wavelength-focus mapping relationship.

[0013] As a preferred solution of the high-precision printed circuit board intelligent visual inspection system of the present invention, wherein: the compensation instruction set based on the wavelength-focus mapping relationship adjusts the imaging light path in real time to generate a high-fidelity fused image, the steps are as follows: The protocol parsing engine identifies the wavelength partition identifier and compensation priority in the compensation instruction set of the wavelength-focus mapping relationship and generates an optical path control parameter package; Based on the optical path control parameter package, the graphene microstructure is driven to perform nanoscale deformation and angular deflection, while simultaneously receiving real-time position feedback to generate a real-time optical path calibration signal stream; The optical path calibration signal stream is used to trigger the acquisition of raw spectral images, and frame accumulation noise reduction and timestamp registration are performed to generate high-fidelity fused images.

[0014] As a preferred solution of the high-precision printed circuit board intelligent visual inspection system of the present invention, the steps of analyzing the changes in optical signal characteristics in the high-fidelity fusion image are as follows: The partitioned histogram equalization algorithm is used to extract optical features from the high-fidelity fused image to enhance the image; An edge-guided region growing algorithm is used to identify brightness mutation areas in optical feature enhanced images, and texture analysis and contour comparison are performed to generate anomaly mask images.

[0015] As a preferred solution of the intelligent visual inspection system for high-precision printed circuit boards of the present invention, the steps of identifying abnormal areas and extracting microscopic defect features to generate an inspection report containing defect features are as follows: According to the abnormal mask image, the geometric feature vector and optical feature vector of the brightness mutation area are extracted, and the defect feature data set is obtained; Based on the defect feature data set, a multi-rule joint decision engine is used to analyze the defect features and generate a detection report containing the defect features.

[0016] The beneficial effects of the present invention are as follows: through the uniform dispersion of quantum dot-modified ink in an inert atmosphere and the plasma substrate activation process, the molecular-level combination of the solder mask and the fluorescent marker material is identified, which serves to convert the stimulated radiation characteristics of the quantum dots into an active defect sensing mechanism; based on the dynamic optical path compensation mechanism of the wavelength-dependent focus shift law, the nanosecond deformation tuning of the graphene microstructure and the real-time alignment of multi-spectral channels are identified, which serves to eliminate the dispersion distortion when the wide-spectrum light source penetrates the multi-layer medium. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 Schematic diagram of the intelligent visual inspection system for high-precision printed circuit boards.

[0019] Figure 2 Flowchart of the quantum dot marking layer preparation process.

[0020] Figure 3 Flowchart generated for the wavelength-focus mapping compensation instruction set.

[0021] Figure 4 Flowchart generated for high-fidelity fused images. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0025] Reference Figures 1 to 4 , is an embodiment of the present invention, which provides a high-precision printed circuit board intelligent visual inspection system, including the following steps: The substrate marking layer coating module builds a uniform fluorescent marking layer on the surface of the printed circuit board through the coating of quantum dot modified ink and light curing process, thus generating a printed circuit board substrate with a quantum dot marking layer attached to the surface; The quantum dot material is mixed with the light-curing solder resist ink in an inert atmosphere, and a modified ink with uniform dispersion of quantum dots is obtained through mechanical stirring and cyclic grinding processes; Furthermore, the quantum dot material and the photocurable solder resist ink are mixed in an inert atmosphere, with mechanical stirring performed by an impeller agitator. The inert atmosphere concentration is maintained stable during the stirring process. The mixed liquid is then transferred to a circulating grinding device for wet grinding. The grinding medium uses a zirconium oxide bead array. The temperature in the grinding chamber is controlled to be constant. The uniformity of the quantum dot material dispersion in the ink is determined by adjusting the bead filling ratio and the rotor speed. Finally, a modified ink with a quantum dot dispersion that meets the standards is output. performing plasma surface activation treatment on a printed circuit board substrate to obtain a surface-activated substrate; Furthermore, the printed circuit board substrate is placed inside a vacuum processing chamber, process gas is introduced under specific pressure conditions, and a high-frequency alternating voltage is applied to the parallel electrode plates to excite the gas to generate low-temperature plasma; the active particles in the plasma bombard the surface of the printed circuit board substrate for a specific period of time, causing the surface molecular chains of the substrate to break, generate free radicals and form oxygen-containing polar groups, and ultimately generate a surface-activated substrate with enhanced ink binding ability.

[0026] Based on the surface activated substrate, the modified ink with uniform dispersion of quantum dots is formed into a wet film coating with uniform thickness through a slot coating process to generate a semi-cured substrate; Furthermore, the surface-activated substrate is fixed in a constant temperature control device, and the modified ink with quantum dot dispersion that meets the standards is input into a precision slot coating device; the gap between the coating scraper and the substrate is adjusted to a set size (determined based on the maximum quantum dot agglomerate size and the substrate surface roughness), and the scraper completes a single stroke at a constant speed to form a wet film coating; the substrate enters a pre-curing zone and receives ultraviolet radiation for a set time (determined based on the ultraviolet absorption coefficient and thermal conductivity coefficient) to achieve initial cross-linking of the coating surface while maintaining a plastic state inside, and finally outputs a semi-cured substrate; Based on the application of staged light irradiation to the semi-cured substrate, the modified ink photo-crosslinking reaction is triggered to form a fully cured substrate, and the quantum dots of the fully cured substrate are simultaneously monitored to generate a printed circuit board substrate with a quantum dot marking layer attached to the surface.

[0027] Furthermore, the semi-cured substrate is subjected to the first stage of light irradiation under a nitrogen protective atmosphere to induce the decomposition of the photoinitiator in the shallow area of ​​the wet film coating to produce free radicals; the second stage of light irradiation is carried out to trigger the photocrosslinking reaction of the ink to completely cure it in the depth direction of the coating; a fiber optic spectrometer is simultaneously used to monitor the fluorescence emission spectrum under the excitation wavelength in real time to confirm that the characteristic fluorescence peak wavelength of the quantum dots is within the set range (determined based on the intrinsic emission spectrum peak of the quantum dot material and the allowable deviation of process fluctuations); when the curing and crosslinking degree reaches the required standard and the fluorescence characteristics of the quantum dots remain stable, the printed circuit board substrate with a quantum dot marking layer attached to the surface is output.

[0028] An image dataset generation module generates an original image dataset by performing confocal scanning using a laser wavelength based on a printed circuit board substrate with a quantum dot marker layer attached to the surface; Based on a printed circuit board substrate with a quantum dot marking layer attached to the surface, the quantum dot marking substrate is fixed by vacuum adsorption, and a motion trajectory solution is generated in combination with machine vision positioning; Furthermore, a printed circuit board substrate with a quantum dot marking layer attached to its surface is placed in the center area of ​​a vacuum adsorption stage, and a vacuum generator is activated to adjust the adsorption air pressure to a set value. An industrial camera photographs the Fiducial marks preset on the surface of the printed circuit board substrate from multiple angles, and extracts the pixel coordinates of the center point of each Fiducial mark. Sub-pixel matching operations are performed on the obtained pixel coordinates with theoretical coordinates in a preset file (generated by computer-aided software based on a circuit wiring diagram before the printed circuit board is manufactured, containing the theoretical position coordinates of all Fiducial marks) to generate a coordinate transformation matrix. Based on the coordinate transformation matrix, the boundary coordinates of the scanning area are obtained, a collision-free path is planned, and a starting point coordinate, a path turning point coordinate sequence, and a scanning speed parameter combination are generated. Finally, a motion trajectory plan including a spatial path and motion parameters is output. Based on the motion trajectory scheme, the laser source is controlled to emit an excitation beam that matches the intrinsic response of the quantum dots, and point-by-point scanning is performed to generate a fluorescence signal stream; Furthermore, based on the starting point coordinates, path turning point sequence, and scanning speed parameters in the motion trajectory scheme, the laser source is controlled to emit an excitation light beam with a wavelength that matches the peak of the quantum dot's intrinsic absorption spectrum. The three-dimensional translation stage moves the laser focus position point by point according to the path turning point sequence, and stays at each position for a set time (determined based on the quantum dot fluorescence lifetime and the signal-to-noise ratio requirements of the detection unit) to stimulate the quantum dot to radiate characteristic fluorescence. The fluorescence signal is detected by a photomultiplier tube, separated from the background noise by a narrow-band filter, and then input into a current-to-voltage conversion circuit. The converted analog voltage signal is collected by an ADC at a set sampling rate, and the spatial coordinate label of the position encoder is synchronously attached to ultimately generate a fluorescence signal stream containing time and space dimensions. It should be noted that the intrinsic response of quantum dots refers to the unique optical behavior exhibited by quantum dot materials under laser excitation of a specific wavelength. Its physical essence stems from the quantum confinement effect and band structure characteristics. When the wavelength of the excitation beam precisely matches the peak of the quantum dot's intrinsic absorption spectrum, the electrons in the quantum dot's valence band are stimulated to transition to the conduction band, forming electron-hole pairs. The excited state of the electron-hole pairs releases energy through radiative recombination, producing characteristic fluorescence with a precise wavelength. The core characteristics are manifested in three dimensions: size dependence results in a positive correlation between the fluorescence wavelength and the quantum dot diameter; surface state regulation ensures material stability through molecular ligand modification; and a defect response mechanism triggers changes in fluorescence intensity in microscopic defect regions, forming an active defect sensing signal source. The defect response mechanism transforms quantum dots into high-precision optical probes. Its value lies in identifying the strict match between the excitation / emission wavelength and the optical device, improving the signal-to-noise ratio through micro-region signal variation, and demonstrating excellent performance in process compatibility, providing a physical basis for breaking through the limits of traditional optical detection.

[0029] The wavelength signal of the fluorescence signal stream is separated by spectral filtering, converted into digital coding information in real time, and a digital matrix of associated spatial positions is generated; Furthermore, the fluorescence signal stream is input into a multi-stage spectral filtering component. The first stage uses a long-pass filter to filter out the residual signal of the excitation wavelength, the second stage uses a band-pass filter to extract the characteristic fluorescence band of the quantum dots, and the third stage uses a short-pass filter to suppress long-wave stray light. The filtered fluorescence signal is converted into a pulse current signal by a photomultiplier tube and then converted into a voltage signal by a transimpedance amplifier. The voltage signal is input into a high-speed ADC analog-to-digital converter for real-time digitization, and the spatial coordinate data output by the position encoder is synchronously collected. The digitized voltage value is bound to the corresponding coordinate and encapsulated into a timestamp-coordinate-grayscale value structure data packet, and finally a digital matrix of associated spatial positions is generated.

[0030] The digital matrix of associated spatial positions is pixel-mapped and reassembled in the trajectory spatial order, and the original image dataset is generated through data integrity verification.

[0031] Furthermore, the digital matrix of associated spatial positions is loaded into the data processing unit, the path turning point sequence of the motion trajectory plan is read as the index sequence, and the data packets in the digital matrix are rearranged according to the traversal rules of ascending X coordinates and ascending Y coordinates. For the gaps between the path turning points in the motion trajectory plan, a bilinear interpolation algorithm is used to fill the grayscale value gaps. The reorganized digital matrix is ​​then subjected to data integrity verification: the cumulative sum of the grayscale values ​​of each row and column is obtained and compared to the preset theoretical total value of the corresponding area in the motion trajectory plan (calculated based on the material reflection coefficient and quantum dot concentration distribution in the substrate design parameters), and the continuity of the data packet timestamp is verified. After verification, the two-dimensional digital matrix is ​​output and encapsulated into an HDF5 format data packet, ultimately generating the original image dataset. It should be noted that pixel mapping and reconstruction is the core processing flow for converting discretely acquired optical signals into spatially ordered images. This process establishes a precise mapping between the laser scanning path and pixels through spatial position reconstruction. Based on the path turning point sequence of the motion trajectory scheme, the digital matrix data is reorganized in ascending XY coordinate order. Bilinear interpolation is used to identify and intelligently fill gaps between path turning points. Rigorous data integrity verification is performed. Finally, a standardized dataset is output through 2D spatial topology reconstruction, HDF5 format packaging, and metadata annotation. This method breaks through the limitations of traditional image reconstruction: adaptive path mapping solves the positioning challenges of irregular scanning trajectories, intelligent gap compensation maintains the geometric fidelity of microstructure edges, and physical constraint verification ensures data reliability, providing a high-precision spatial topology data foundation for defect identification.

[0032] The focus shift compensation module performs multispectral feature analysis on the original image data set, captures the focus shift patterns of light of different wavelengths, and generates a wavelength-focus mapping relationship compensation instruction set; Based on the original image data set, a multi-spectral separation algorithm is used to extract the fluorescence wavelength channel and the excitation wavelength residual channel, and spatial frequency domain analysis is performed to identify the feature point coordinate set; More specifically, the original image data set is input into the multi-spectral separation algorithm processing unit, and the blind source separation algorithm is applied to decompose the mixed spectral signal to extract the quantum dot characteristic fluorescence wavelength channel and the excitation wavelength residual channel; a two-dimensional fast Fourier transform is performed on the separated wavelength channel images to generate a frequency domain spectrum, and the spatial frequency domain features are extracted through a bandpass frequency filter; the energy concentration area in the frequency domain spectrum is identified as a feature candidate point, the coordinates of the center of gravity of the candidate point are obtained and mapped back to the spatial domain position, and a coordinate pair set including the coordinates of the fluorescence wavelength channel feature point and the coordinates of the excitation wavelength channel feature point is generated, and finally the feature point coordinate set is output; It should be noted that the multispectral unmixing algorithm is the core processing step for achieving precise optical feature extraction. It decomposes the mixed optical signal in the original image into independent spectral components through a blind source separation computational framework, and unmixes the signal by utilizing the intrinsic wavelength difference between the characteristic fluorescence of quantum dots and the residual excitation light. The multispectral unmixing algorithm mimics the spectral unmixing mechanism of cone cells in the human eye but transcends biological limitations, achieving nanometer-level wavelength resolution through independent component analysis. Its core innovative advantages include adaptive unmixing capabilities, frequency domain feature enhancement, and spatial positioning accuracy. It successfully resolves the spectral aliasing problem of broadband light sources penetrating multilayer media, ensuring signal reliability in complex noisy environments. It also improves the positioning accuracy of micron-level feature points to the nanometer level, providing a precise spatial reference for high-density PCB inspection and raising the recognition rate of inner-layer defects to industry-leading levels.

[0033] Calculate the spatial position offset vector of the same feature point in the feature point coordinate set under the fluorescence wavelength channel and the excitation wavelength channel to generate a wavelength-focus offset characteristic curve; More specifically, a correspondence is established based on the fluorescence wavelength channel feature points and the excitation wavelength channel feature points in the feature point coordinate set according to the spatial Euclidean distance matching rule. When the straight-line distance between the two points does not exceed the set maximum tolerance distance (set based on the maximum feature point position deviation allowed by the substrate manufacturing process and the spatial resolution limit of the detection system), it is recorded as a valid coordinate pair; a coordinate difference operation is performed on each valid coordinate pair: the horizontal offset is obtained by subtracting the X coordinate of the fluorescence channel feature point from the X coordinate of the excitation channel feature point, and the vertical offset is obtained by subtracting the Y coordinate of the fluorescence channel feature point from the Y coordinate of the excitation channel feature point; all offset data are grouped by wavelength range, and the arithmetic mean of the horizontal offset and vertical offset within the same group is calculated; the variation pattern of the wavelength value and the offset mean is established through data fitting, and finally a wavelength-focus offset characteristic curve is output and stored in a JSON file format, including a wavelength-horizontal offset mapping table, a wavelength-vertical offset mapping table, and statistical analysis confidence interval parameters; Spatial position offset and characteristic curve generation formula: ; in, represents the group average offset vector; Indicates the The number of valid coordinate pairs of the group; Represents a point pair index; Indicates the For the characteristic points of the fluorescence channel coordinate; Indicates the For the characteristic points of the excitation channel coordinate; Indicates the For the characteristic points of the fluorescence channel coordinate; Indicates the For the characteristic points of the excitation channel coordinate; represents the excitation channel; Indicates fluorescence channels.

[0034] According to the wavelength-focus shift characteristic curve, the focus compensation required for each wavelength is calculated, and the compensation sequence is optimized using the gradient descent method to generate a compensation instruction set of the wavelength-focus mapping relationship.

[0035] More specifically, based on the wavelength-horizontal offset mapping table and the wavelength-vertical offset mapping table in the wavelength-focus offset characteristic curve, the horizontal focus compensation amount and the vertical focus compensation amount corresponding to each wavelength node are extracted in ascending order of wavelength; the compensation amount sequence is input into the iterative optimization algorithm, and the compensation amount value is adjusted through continuous iteration so that the compensation error function value is gradually reduced, thereby generating an optimized focus compensation sequence for each wavelength; the compensation sequence is encapsulated into a binary protocol, and finally a compensation instruction set of the wavelength-focus mapping relationship is output; ; in, represents the compensation instruction set; Indicates the The compensation amount of each wavelength node; Indicates the The compensation amount of each wavelength node; represents the wavelength interval; represents the fidelity weight; Indicates the number of wavelength points; Indicates the The estimated value of the initial compensation amount corresponding to the wavelength node; The imaging optical path adjustment module adjusts the imaging optical path in real time based on the compensation instruction set of the wavelength-focus mapping relationship to generate a high-fidelity fused image; The protocol parsing engine identifies the wavelength partition identifier and compensation priority in the compensation instruction set of the wavelength-focus mapping relationship and generates an optical path control parameter package; More specifically, the compensation instruction set of the wavelength-focus mapping relationship is input into the protocol parsing engine, which first parses the four bytes of the protocol header to obtain the wavelength partition identifier; then reads the priority tag field; decodes the compensation data body in the order of the wavelength partition, and extracts the axial compensation and radial compensation corresponding to each wavelength node; converts the compensation parameters into voltage control instructions, and finally encapsulates them into a structured optical path control parameter package.

[0036] Based on the optical path control parameter package, the graphene microstructure is driven to perform nanoscale deformation and angular deflection, while simultaneously receiving real-time position feedback to generate a real-time optical path calibration signal stream. More specifically, the angle of the graphene nanounit is deflected through the interdigitated electrode array according to the radial voltage command; the real-time position feedback signal of the piezoelectric micro-displacement device is synchronously received, and the deformation depth error and angular deflection error are obtained using a proportional-integral-differential closed-loop control algorithm; the real-time deformation depth value, deflection angle value, and calibration status identifier are encapsulated into a data stream, and finally a real-time optical path calibration signal stream is generated; It should be noted that the radial voltage command is derived from the axial displacement parameter in the wavelength-focus mapping relationship compensation command set and is obtained by analyzing the voltage-displacement conversion function; It should be explained that the graphene microstructure is a micro-nanoscale optical element composed of a single layer of carbon atoms, and the honeycomb lattice structure realizes real-time control of the optical path through voltage control: when the radial voltage is applied to the interdigitated electrode, the π bond electron cloud is rearranged, resulting in a high-precision deformation response (in-plane strain causes the carbon atom bond length to be dynamically adjusted, and out-of-plane bending changes the hybrid orbital angle); the asymmetric electrode design establishes a voltage gradient to induce non-uniform stress in the lattice, realizing high-speed deflection control over a large angle range; forming a closed loop with the piezoelectric micro-displacement device, optical monitoring tracks the deformation depth and deflection angle in real time; the graphene microstructure has a wide spectrum control capability, realizing dynamic adjustment of the refractive index, tuning of plasma characteristics and precise compensation of the optical path, breaking through the limitations of slow response speed and short service life of traditional optical components, and realizing high-precision axial chromatic aberration compensation in multi-spectral imaging to meet the needs of precision electronic manufacturing and testing.

[0037] Use the optical path calibration signal stream to trigger the acquisition of raw spectral images, perform frame accumulation noise reduction and timestamp registration, and generate high-fidelity fused images; More specifically, the core features of the unprocessed optical data acquired by the multi-channel synchronous acquisition device include optical information from three independent channels: UV channel data, captured by a back-illuminated sCMOS detector, primarily includes quantum dot fluorescence characteristics and microcrack scattering signals; visible channel data, acquired by a global shutter CMOS detector, records substrate surface topography and solder joint reflectance characteristics; and near-infrared channel data, acquired by an InGaAs array detector, penetrates multilayer dielectrics to reveal internal structural information. Each channel undergoes pixel-level frame accumulation, and spatiotemporal synchronization is achieved through timestamp registration. An adaptive weighting algorithm dynamically adjusts the number of accumulated frames based on the signal-to-noise ratio (SNR). Frames are reduced in high SNR areas to improve efficiency, while frames are increased in low SNR areas to enhance detail. The registered 3D dataset forms a raw spectral image cube. The spatial dimension records the physical coordinates of the substrate, the spectral dimension carries wavelength characteristics, and the temporal dimension preserves dynamic process data. Subsequent processing uses a wavelength-focus mapping compensation instruction set to spatially align multi-channel feature points to eliminate optical dispersion distortion. When combining spectral channels using a weighted pyramid fusion algorithm, the fusion weights are adaptively adjusted based on the distribution of quantum dot fluorescence intensity: the UV channel enhances microcrack features (weight boost), the visible channel preserves surface texture (baseline weight), and the near-infrared channel suppresses background noise (weight decay). After fusion, edge-adaptive sharpening is performed to compensate for optical diffraction effects, and the image is finally converted to the sRGB color space to produce a high-fidelity image that conforms to human visual characteristics.

[0038] The defect report generation module analyzes the changes in optical signal characteristics in high-fidelity fusion images, identifies abnormal areas and extracts microscopic defect features, and generates an inspection report containing defect features.

[0039] The partitioned histogram equalization algorithm is used to extract optical features from the high-fidelity fused image to enhance the image; Furthermore, the high-fidelity fusion image input processing unit divides the image into fixed-size blocks; performs histogram statistics on the grayscale values ​​of pixels in each block and obtains the grayscale cumulative distribution function of the block; establishes a grayscale mapping relationship based on the cumulative distribution function, maps the original grayscale value to the target grayscale range (set by analyzing the cumulative distribution function of the quantum dot fluorescence intensity), and implements contrast suppression on the quantum dot aggregation area during the execution process; reorganizes the pixel values ​​after processing of all blocks, performs bilinear interpolation smoothing on the block boundaries, and finally outputs an optical feature enhanced image.

[0040] An edge-guided region growing algorithm is used to identify brightness mutation areas in optical feature enhanced images, and texture analysis and contour comparison are performed to generate anomaly mask images. Furthermore, the optical feature enhanced image input processing unit uses an edge detection algorithm to extract the image gradient amplitude map, and uses the gradient maximum point as the region growth seed point; when the difference between the pixel grayscale value and the grayscale of the growth seed point is less than a set percentage and the texture uniformity index dual criteria, region growth is performed to generate a set of connected regions; morphological contour polygons are extracted for each connected region, and distance matching is performed with the standard component contour polygons in the substrate file, and polygonal regions whose matching distance exceeds the tolerance range (set jointly according to the position tolerance band width of the standard component contour in the substrate file and the extraction accuracy of the feature point coordinate set) are marked as abnormal areas; and finally, an abnormal mask image is output.

[0041] It should be noted that the abnormal mask image is a high-precision binary feature-marked image generated through edge-guided region growth and intelligent contour comparison. The abnormal mask image locates seed points based on the gradient information of the optical feature enhancement image. It dynamically expands the connected region using a grayscale difference threshold (dynamically generated by analyzing the grayscale statistical characteristics of the local neighborhood of the component under test) and texture uniformity as dual criteria to form a complete abnormal region boundary. A morphological contour extraction algorithm is used to convert the connected region into a polygonal structure, which is then intelligently matched with the standard component contours in the standard file. Ultimately, areas outside the tolerance range are marked as abnormal regions, and an abnormal mask image is output. The abnormal mask image has sub-pixel edge positioning capabilities, fully preserving the true topological structure of micro-defects. It uses black and white binary values ​​to intuitively distinguish normal and abnormal regions. Its spatial accuracy supports micron-level defect recognition, its geometric fidelity avoids over-segmentation misjudgments, and its dynamic tolerance control adapts to process fluctuations, improving the accuracy of identifying cold solder joints and microcracks in precision electronics manufacturing inspections.

[0042] According to the abnormal mask image, the geometric feature vector and optical feature vector of the brightness mutation area are extracted, and the defect feature data set is obtained; The abnormal mask image is input into the feature extraction unit, which first executes the connected domain labeling algorithm to identify the independent contours of each brightness mutation area. For each independent connected domain, the geometric feature vector is extracted, including the equivalent diameter, aspect ratio, and contour circularity. Simultaneously, the optical feature vector is extracted from the high-fidelity fused image at the corresponding position, including the maximum light intensity gradient value, average fluorescence intensity, and texture uniformity. All features are organized into structured data records according to the connected domain identification number. Finally, the data is packaged into a defect feature dataset containing the feature vectors of all abnormal regions. Based on the defect feature data set, a multi-rule joint decision engine is used to analyze the defect features and generate a detection report containing the defect features.

[0043] The defect feature data set is input into the multi-rule joint decision-making engine. First, the preset defect classification parameter table is called (by statistically analyzing the distribution range of geometric and optical features of historical defect samples on the production line, and combining destructive physical verification experiments to determine the quantitative interval boundary values ​​of each defect type), and a quantitative judgment is performed on the geometric feature vector of each data record; the optical feature vector is analyzed simultaneously; the position coordinates of the successfully matched defects are mapped, and feature priority arbitration is performed when a single defect triggers a multi-rule conflict; an inspection report containing three types of core data is generated: a defect location heat map, a type ratio distribution statistical map, and a size grade classification table; and finally, a formatted inspection report containing defect features is output.

[0044] In summary, the present invention achieves molecular-level bonding between the solder mask and the fluorescent marker material through: uniform dispersion of quantum dot-modified ink in an inert atmosphere and plasma substrate activation process, which converts the stimulated emission characteristics of quantum dots into an active defect sensing mechanism; a dynamic optical path compensation mechanism based on the wavelength-dependent focus shift law achieves nanosecond-level deformation tuning of the graphene microstructure and real-time alignment of multi-spectral channels, which eliminates dispersion distortion when a wide-spectrum light source penetrates a multi-layer medium.

[0045] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A high-precision printed circuit board intelligent visual inspection system, characterized by: include, The substrate marking layer coating module builds a uniform fluorescent marking layer on the surface of the printed circuit board through the coating of quantum dot modified ink and light curing process, thus generating a printed circuit board substrate with a quantum dot marking layer attached to the surface; An image dataset generation module generates an original image dataset by performing confocal scanning using a laser wavelength based on a printed circuit board substrate with a quantum dot marker layer attached to the surface; The focus shift compensation module performs multispectral feature analysis on the original image data set, captures the focus shift patterns of light of different wavelengths, and generates a wavelength-focus mapping relationship compensation instruction set; The imaging optical path adjustment module adjusts the imaging optical path in real time based on the compensation instruction set of the wavelength-focus mapping relationship to generate a high-fidelity fused image; The defect report generation module analyzes the changes in optical signal characteristics in high-fidelity fusion images, identifies abnormal areas and extracts microscopic defect features, and generates an inspection report containing defect features.

2. The high-precision printed circuit board intelligent visual inspection system according to claim 1, characterized in that: The steps of the quantum dot modified ink coating and light curing process are as follows: The quantum dot material is mixed with the light-curing solder resist ink in an inert atmosphere, and a modified ink with uniform dispersion of quantum dots is obtained through mechanical stirring and cyclic grinding processes; The printed circuit board substrate is subjected to plasma surface activation treatment to obtain a surface activated substrate.

3. The high-precision printed circuit board intelligent visual inspection system according to claim 2, characterized in that: The steps of constructing a uniform fluorescent marking layer on the surface of the printed circuit board to generate a printed circuit board substrate with a quantum dot marking layer attached to the surface are as follows: Based on the surface activated substrate, the modified ink with uniform dispersion of quantum dots is formed into a wet film coating with uniform thickness through a slot coating process to generate a semi-cured substrate; Based on the application of staged light irradiation to the semi-cured substrate, the modified ink photo-crosslinking reaction is triggered to form a fully cured substrate, and the quantum dots of the fully cured substrate are simultaneously monitored to generate a printed circuit board substrate with a quantum dot marking layer attached to the surface.

4. The high-precision printed circuit board intelligent visual inspection system according to claim 3, characterized in that: The printed circuit board substrate based on the surface-attached quantum dot marking layer comprises the following steps: Based on a printed circuit board substrate with a quantum dot marking layer attached to the surface, the quantum dot marking substrate is fixed by vacuum adsorption, and a motion trajectory solution is generated in combination with machine vision positioning; Based on the motion trajectory scheme, the laser source is controlled to emit an excitation beam that matches the intrinsic response of the quantum dots, and point-by-point scanning is performed to generate a fluorescence signal stream.

5. The high-precision printed circuit board intelligent visual inspection system according to claim 4, characterized in that: The laser wavelength is used for confocal scanning to generate the original image data set, and the steps are as follows: The wavelength signal of the fluorescence signal stream is separated by spectral filtering, converted into digital coding information in real time, and a digital matrix of associated spatial positions is generated; The digital matrix of associated spatial positions is pixel-mapped and reassembled in the trajectory spatial order, and the original image dataset is generated through data integrity verification.

6. The high-precision printed circuit board intelligent visual inspection system according to claim 5, characterized in that: The multispectral feature analysis based on the original image data set refers to using a multispectral separation algorithm to extract the fluorescence wavelength channel and the excitation wavelength residual channel, and performing spatial frequency domain analysis to identify the feature point coordinate set.

7. The high-precision printed circuit board intelligent visual inspection system according to claim 6, characterized in that: The steps of capturing the focus shift pattern of light of different wavelengths and generating a wavelength-focus mapping relationship compensation instruction set are as follows: Calculate the spatial position offset vector of the same feature point in the feature point coordinate set under the fluorescence wavelength channel and the excitation wavelength channel to generate a wavelength-focus offset characteristic curve; According to the wavelength-focus shift characteristic curve, the focus compensation required for each wavelength is calculated, and the compensation sequence is optimized using the gradient descent method to generate a compensation instruction set of the wavelength-focus mapping relationship.

8. The high-precision printed circuit board intelligent visual inspection system according to claim 7, characterized in that: The compensation instruction set based on the wavelength-focus mapping relationship adjusts the imaging light path in real time to generate a high-fidelity fused image. The steps are as follows: The protocol parsing engine identifies the wavelength partition identifier and compensation priority in the compensation instruction set of the wavelength-focus mapping relationship and generates an optical path control parameter package; Based on the optical path control parameter package, the graphene microstructure is driven to perform nanoscale deformation and angular deflection, while simultaneously receiving real-time position feedback to generate a real-time optical path calibration signal stream; The optical path calibration signal stream is used to trigger the acquisition of raw spectral images, and frame accumulation noise reduction and timestamp registration are performed to generate high-fidelity fused images.

9. The high-precision printed circuit board intelligent visual inspection system according to claim 8, characterized in that: The steps of analyzing the changes in optical signal characteristics in the high-fidelity fusion image are as follows: The partitioned histogram equalization algorithm is used to extract optical features from the high-fidelity fused image to enhance the image; An edge-guided region growing algorithm is used to identify brightness mutation areas in optical feature enhanced images, and texture analysis and contour comparison are performed to generate anomaly mask images.

10. The high-precision printed circuit board intelligent visual inspection system according to claim 9, characterized in that: The steps of identifying abnormal areas and extracting microscopic defect features to generate a test report containing defect features are as follows: According to the abnormal mask image, the geometric feature vector and optical feature vector of the brightness mutation area are extracted, and the defect feature data set is obtained; Based on the defect feature data set, a multi-rule joint decision engine is used to analyze the defect features and generate a detection report containing the defect features.

Images