59 results about "Anisotropic diffusion filtering" patented technology

The invention relates to the technical field of image data processing, in particular to a PCB soldering paste printing quality detection method and system based on image processing, and the method comprises the steps: collecting a surface image of a to-be-detected PCB; performing filtering processing on the surface image of the PCB to be detected by using an improved anisotropic diffusion filtering algorithm to obtain a filtered surface image of the PCB, and performing solder paste offset defect identification on the filtered surface image of the PCB based on a convolutional neural network; wherein the improved anisotropic diffusion filtering algorithm comprises a K value, the K value is a product of an initial K value and a correction coefficient, and the correction coefficient is in positive correlation with the noise intensity and the importance degree of each pixel point in the PCB surface image. The problem that an existing filtering algorithm is not high in detection accuracy is solved.

The invention provides a scanning electron microscope image edge detection method, which comprises the following steps: carrying out anisotropic diffusion filtering on a scanning electron microscope image to suppress noise and reserve edges to obtain a filtered image; an original scale gradient and a down-sampling scale gradient of the filtered image are calculated, a plurality of pixel regions in the original scale gradient and the down-sampling scale gradient are fused based on a plurality of adaptive weights to obtain a gradient map, the adaptive weights are determined based on the complexity of the texture of the filtered image, and each pixel region corresponds to one adaptive weight; multi-dimensional features are extracted from the gradient map, the multi-dimensional features are input into a machine learning model for threshold prediction, a continuous pixel-level threshold map is generated based on a predicted threshold, and a gradient threshold included in the continuous pixel-level threshold map is a critical value for distinguishing different types of pixels in the gradient map; and comparing the gradient map with the continuous pixel-level threshold map to obtain a target edge pixel, and generating an edge detection result map based on the target edge pixel.

The invention provides a multi-parameter feature fused vertebral CT image standardization preprocessing method and system. The method comprises the following steps: carrying out initial segmentation on an input original vertebral CT image; carrying out enhancement processing on the image by adopting an anisotropic diffusion filtering algorithm of multi-scale structure perception; establishing a tissue specificity density mapping model of multi-parameter fusion; adopting a self-adaptive segmentation histogram mapping method to map gray distribution of an original image to a standard template; and constructing a rigid-elastic mixed registration model. According to the invention, a complete quality evaluation system is established, the processing quality can be automatically detected and reported, and the reliability of an output result is ensured. By introducing a patient individualized correction mechanism, individual difference information is fully reserved while the standardization effect is ensured, and diagnosis information loss caused by excessive standardization is avoided. The multi-parameter fusion strategy ensures complete reservation of bone tissue density, texture and structure information.

The invention relates to the technical field of image data processing, in particular to a real-time denoising method and system for a linear guide rail scanning image, and the method comprises the steps: obtaining a to-be-processed linear guide rail scanning image; and de-noising the to-be-processed linear guide rail scanning image by using the improved anisotropic diffusion filtering algorithm to obtain a de-noised linear guide rail scanning image. Wherein the improved anisotropic diffusion filtering algorithm comprises a diffusion function, the diffusion function is in positive correlation with an initial diffusion function and the artifact boundary confidence coefficient of each pixel point, and the artifact boundary confidence coefficient represents the confidence coefficient of each pixel point belonging to an artifact. The problem that an existing algorithm is insufficient in adaptability when facing complex noise is solved.

The invention provides a multi-arm spiral subarray cooperative adaptive acoustic imaging system and method. The system adopts a circular substrate to construct an Archimedes spiral array. The innovative dynamic processing strategy comprises low-frequency sub-array combination and aperture expansion, intermediate-frequency independent processing and load balance maintenance, and high-frequency cross-arm virtual focusing array construction. The key technology covers multi-channel dynamic sampling, frequency band conflict arbitration, phase compensation beam forming and confidence weighted fusion. In the data fusion stage, a blind area is eliminated by adopting Delaunay triangulation and radial basis interpolation, and edge features are reserved in combination with anisotropic diffusion filtering. A temperature-sound velocity correction model is integrated in the system, and delay summation, MVDR and compressed sensing hybrid beam forming algorithms are applied to real-time processing. The output unit generates a three-dimensional acoustic thermodynamic diagram fusing Gaussian smoothing and edge enhancement. According to the scheme, the spatial sampling efficiency is optimized through spiral topology, the resolution and precision in a complex scene are improved through a dynamic recombination mechanism, and the method is suitable for industrial monitoring and noise source positioning.

The invention relates to an anti-NMDAR encephalitis clinical prognosis evaluation method based on artificial intelligence, and belongs to the technical field of artificial intelligence and data processing. The method comprises the following steps: acquiring multi-modal nerve image data of a patient; performing fusion preprocessing on the multi-modal neural image data by adopting a tensor decomposition fusion strategy, and reserving cross-modal spatial correlation through low-rank constraint to obtain a fused output tensor; carrying out focus perception anisotropic diffusion filtering on the fused output tensor to obtain an output image after diffusion filtering; an anti-NMDAR encephalitis clinical prognosis evaluation model is constructed, the output image after diffusion filtering is input into the model for training, an Adam adaptive optimizer is adopted to optimize the training process, and finally a trained model is obtained; and inputting a to-be-evaluated output image after diffusion filtering into the trained model to obtain an evaluation classification result. The identification and classification capability of the model on the focus can be enhanced.

The invention discloses an earthquake intelligent filtering method based on multi-scale information fusion and structural feature enhancement, and belongs to the technical field of geophysical exploration data processing, and the method comprises the following steps: obtaining original earthquake data, carrying out the normalization processing of the data, and obtaining the preprocessing earthquake data; performing anisotropic diffusion filtering processing on the original seismic data based on an anisotropic diffusion equation to obtain structure enhanced seismic data; constructing a parallel convolutional neural network, and performing feature extraction on the preprocessed seismic data and the structure enhanced seismic data to obtain a multi-scale seismic feature image; based on a feature fusion neural network, performing cross-scale fusion on the multi-scale feature image to generate a fused feature image; and constructing and training a residual neural network MA-ResCNN based on multi-scale fusion and an attention mechanism. The method can effectively remove the noise interference of the exploration seismic data in a complex structure region, and improves the resolution and the signal-to-noise ratio of the seismic data.

The invention discloses an AI-based ultrasonic image feature extraction and recognition system and method, and particularly relates to the field of image feature recognition, and the system comprises a bimodal imaging module, an image preprocessing and registration fusion module, an AI feature extraction module, an AI classification and recognition module, and a diagnosis report generation module. The method comprises the following steps: constructing a bimodal image data set by synchronously collecting ultrasonic and photoacoustic information; after anisotropic diffusion filtering and photon inversion reconstruction preprocessing are adopted, sub-pixel-level space alignment is realized based on B-spline non-rigid registration, and multi-scale fusion is completed by using a Laplacian pyramid; designing an improved DenseNet-121 network to extract form, texture, blood flow and functional four-dimensional feature vectors, inputting the feature vectors into an integrated classifier after L1 regularization dimensionality reduction, and outputting benign and malignant probabilities through a dynamic weighting strategy; clinical priori knowledge is combined to calibrate confidence, and a structured report containing three-dimensional positioning, BI-RADS grading and risk prompting is automatically generated.

The invention relates to the technical field of medical image detection, in particular to a medical image analysis method and system based on image processing, and the method comprises the following steps: collecting a diffusion tensor image, carrying out the filtering preprocessing, normalizing the main diffusion direction, extracting an energy super-threshold voxel, calculating the node distance offset, and carrying out the correlation analysis to generate an atlas. And interpolating and aligning the brain return boundary to obtain a coupling graph, and outputting an abnormal probability distribution graph in a classified manner. According to the method, fiber details are ensured by adopting anisotropic diffusion filtering noise reduction of a diffusion tensor image sequence, three-dimensional normalization and wavelet energy analysis are coordinated, node detection specificity is enhanced by dynamic threshold screening, space-time coupling is quantified by combining displacement Euclidean distance and time sequence relevance, and an individualized coupling model is established by interpolating and fusing brain return boundaries; the template rigid constraint is broken through, the support vector machine recognizes an abnormal mode, structural distortion and functional abnormality cross-scale detection is achieved, and the early lesion image marker recognition efficiency is improved.

The invention relates to the technical field of image recognition, and discloses an industrial pipeline welding defect recognition system based on a DR detection image, and the system comprises the steps: obtaining geometric parameters through a laser contour scanner, calculating a radial distortion amount, and carrying out the curved surface distortion correction of an industrial pipeline; exposure parameters are obtained, a welding seam type and exposure parameter mapping table is constructed, an amplitude value is obtained, and the edge diffusion amount is reduced; after the edge diffusion amount is reduced, a scattered field model of a multi-layer medium is constructed, an effective gray value is calculated, anisotropic diffusion filtering is combined to enhance welding seam area features, and material difference correction and gray offset compensation correction are carried out on dissimilar steel welding; establishing a double-branch decoupling network, converting a welding seam path into a welding seam binary template, calculating fusion line constraint loss, and training an enhancement strategy by adopting a mixed data set; constructing an exposure parameter transfer learning model, and normalizing the cross-working-condition image through a gray transfer function; and constructing standard rule libraries, and calling multiple standard rule libraries in real time to carry out compliance judgment.

The invention belongs to the technical field of condition judgment, and particularly relates to an electrode foil corrosion condition judgment method and system, and the method comprises the following steps: S1, obtaining a grayscale image of a to-be-analyzed electrode foil; constructing a structure tensor based on the composite gradient reflecting the local brightness and texture information of the pixel points, determining an anisotropic diffusion coefficient according to the structure tensor, and performing iterative anisotropic diffusion filtering processing on the grayscale image by using the anisotropic diffusion coefficient to obtain a filtered image; and S2, fusing the anisotropic diffusion coefficient determined for each pixel point with the pixel intensity of the filtered image to obtain a high-contrast corrosion significance map. According to the method, the contour edge information of the corrosion area can be kept while the complex texture and noise of the background area of the electrode foil image are smoothed, the contradiction between denoising and edge protection of a traditional filtering method is solved, the result of the corrosion condition is more reliable, and the automation level of electrode foil product quality detection is improved.

The invention discloses OLED defect detection equipment based on machine vision, and relates to the technical field of visual detection, and the detection equipment comprises the following working processes: driving an OLED panel through a sinusoidal modulation current sequence, synchronously collecting a plurality of light-emitting images, and carrying out registration; constructing a three-dimensional texture tensor and generating a scale weighted texture field through wavelet decomposition; the local entropy density and the multi-scale gradient are fused, and a defect sensitive entropy change field is calculated in combination with the direction constraint matrix; non-linearly enhancing the defect area, and executing anisotropic diffusion filtering to optimize the signal-to-noise ratio; extracting a candidate region, and iteratively solving an energy functional based on an improved Snake model to realize sub-pixel boundary fitting; calculating morphological and optical characteristic parameters such as a defect area, a contrast ratio and a spectrum abnormal index; and performing decision tree classification on the feature vectors, identifying defect types, and generating a report. According to the method, multi-scale texture response of the OLED panel is excited through active current modulation, and electromagnetic feature expression of microdefects is enhanced in combination with an entropy change field algorithm.

The invention provides a seismic fault identification method and system based on multi-attribute collaboration and spectral domain graph enhancement, and relates to the technical field of seismic exploration, and the method comprises the steps: carrying out the all-directional dip angle and azimuth angle scanning of a seismic data volume, extracting multi-scale guide field information through combining structure tensor decomposition, and executing anisotropic diffusion filtering; calculating characteristic value distribution through a characteristic value coherence algorithm based on the filtering data volume, and determining the structural consistency difference of adjacent seismic traces to obtain a fracture coherence attribute data volume; carrying out azimuth gather sorting and pre-stack time migration processing on the seismic data volume, extracting seismic wave dynamic response characteristics, carrying out Fourier series expansion on the azimuth change rate to obtain a crack indication information data volume, splicing the data volume and executing multi-scale three-dimensional convolution solution, and constructing a spatial dependency graph through a spectral clustering algorithm; spectral domain enhancement features are obtained through spectral domain graph transformation and frequency selective filtering reconstruction, and morphological connectivity analysis is executed to obtain a crack prediction result.

This invention discloses a method and system for following up on magnetic resonance imaging (MRI) images of amyloid-related radiological abnormalities, belonging to the field of medical image processing. The method includes: acquiring multimodal MRI image data; performing bias field correction and intensity normalization based on a bias field fitting model to generate a first image set; performing global alignment and edge detection on the first image set to extract anatomical physical boundaries, generating brain region mask data; performing branch deformation field registration on the first image set within the brain parenchyma defined by the brain region mask data to generate a spatially aligned second image set; calculating the average intensity of normal brain parenchyma regions in the second image set to obtain a scaling factor, acquiring a three-dimensional residual image; performing anisotropic diffusion filtering denoising on the three-dimensional residual image, and performing differential denoising and morphological screening processing, outputting the detection results. This invention eliminates systematic artifacts caused by insufficient bias field correction.

This invention relates to the field of image data processing technology, specifically to a method and system for detecting the edge polishing quality of mobile phone glass covers. The method includes: capturing an image of the mobile phone glass cover to be inspected; filtering the image using an improved anisotropic diffusion filtering algorithm; and performing edge detection on the filtered image to obtain multiple edge contours. If the number of edge contours exceeds a set threshold, the image is deemed to be of substandard quality. This invention solves the problem of low accuracy in edge polishing quality detection of mobile phone glass covers using existing anisotropic diffusion filtering algorithms.

The invention relates to the technical field of medical image processing, and discloses a chronic obstructive pulmonary emphysema distribution quantitative method and system based on multi-scale segmentation, and the method comprises the steps: carrying out the anisotropic diffusion filtering noise reduction of a chest CT image; segmenting a lung field and removing a blood vessel bronchial structure by adopting a region growing algorithm; multi-scale image representation is constructed based on a Gaussian pyramid, an emphysema candidate area is identified in a coarse scale layer, and a boundary is accurately drawn by adopting a self-adaptive threshold value in a fine scale layer; extracting local texture features to distinguish the lobular central emphysema and the total lobular emphysema; dividing severity levels according to spatial aggregation characteristics and density gradient distribution, and calculating an air swelling volume ratio and a distribution heterogeneity index; the three-dimensional pseudo-color volume is used for drawing visualization, a structured quantitative report is generated, accurate segmentation and subtype classification of the emphysema area are achieved, and comprehensive quantitative analysis indexes are provided.

This invention relates to the field of image processing technology and discloses a multimodal data fusion method for online quality monitoring of plastics. The method includes: simultaneously acquiring a reference optical image of the plastic surface and a temporal infrared thermal distribution image containing the current frame and the previous frame infrared image; calculating a spatial temperature gradient matrix; correcting the geometric affine matrix based on the displacement vector normal component determined by the dense optical flow field, constructing a spatiotemporal compensation mapping matrix to achieve alignment of heterogeneous pixel coordinate systems, and generating a structure-guided tensor; applying morphological opening operations to remove low-frequency background thermal fluctuations from the structure-guided tensor to obtain a pure-state high-frequency thermal gradient matrix; adjusting the diffusion coefficient using the pure-state high-frequency thermal gradient matrix as a constraint, applying anisotropic diffusion filtering to the reference optical image, and extracting defect regions. This invention achieves physical-level decoupling between optical artifacts and real defects, eliminates feature mapping deviations caused by thermal conduction hysteresis, and improves the signal-to-noise ratio of image feature extraction.

The present application belongs to the technical field of image data processing, and particularly relates to a method for analyzing pressure uniformity of an impression roller based on indentation image features, which comprises the following steps: obtaining an indentation grayscale image; determining local texture complexity of a pixel point; determining neighborhood gradient direction consistency and real noise disturbance index of the pixel point; determining an adaptive gradient threshold value of the pixel point; obtaining a filtered indentation grayscale image, extracting indentation width features, and evaluating pressure uniformity of the impression roller. The present application analyzes the gray value features and gradient direction features of the pixel points in the neighborhood of the pixel point, extracts the indentation under the interference of wear texture, overcomes the difficulty of distinguishing the wear texture and the indentation by using a fixed gradient threshold value in the traditional anisotropic diffusion filtering algorithm, filters out the wear texture noise interference and extracts the indentation by using the adaptive gradient threshold value, and improves the accuracy of the evaluation of the pressure uniformity of the impression roller.

This invention relates to the field of image data processing technology, specifically to a method and system for detecting solder paste printing quality on PCB boards based on image processing. The method includes: acquiring an image of the surface of a PCB board to be inspected; filtering the image using an improved anisotropic diffusion filtering algorithm to obtain a filtered image of the PCB board surface; and identifying solder paste misalignment defects in the filtered image using a convolutional neural network. The improved anisotropic diffusion filtering algorithm includes a K value, which is the product of an initial K value and a correction coefficient. The correction coefficient is positively correlated with the noise intensity and importance of each pixel in the PCB board surface image. This invention solves the problem of low detection accuracy in existing filtering algorithms.

The invention discloses a marine sandstone inter-gravel pore semantic segmentation method based on multi-model collaboration, and relates to the field of marine sandstone inter-gravel pore image segmentation, and the method comprises the steps: obtaining multi-source marine sandstone data, carrying out format analysis, metadata extraction, gray level normalization and size unification, and outputting standardized data; performing anisotropic diffusion filtering, directional gradient enhancement and channel-space joint noise sensing on the standardized data to generate intermediate representation with a high signal-to-noise ratio; respectively inputting the intermediate representation into three model branches of U-Net, DeepLabV < 3 + > and PSPNet for collaborative segmentation, and outputting a plurality of model features; performing weighted fusion on the plurality of model features based on an attention mechanism to obtain fused features; and performing inter-gravel pore repair and artifact suppression on the fused features, and outputting a final segmentation result. The method has the beneficial effects that the precision, robustness and practicability of the scheme are improved through innovative technologies such as multi-model collaboration, dynamic fusion, noise suppression and adaptive receptive field adjustment.

The invention discloses a PCB identifying and grabbing method and system based on machine vision. The method comprises the steps that an original image of a target PCB on a conveying belt is acquired; performing adaptive Gaussian filtering and anisotropic diffusion filtering based on whale swarm algorithm control on the original image to obtain a second smooth image; calculating the second smooth image by using a de-Gaussian filtering canny algorithm to obtain a contour image; acquiring an affine transformation matrix of the contour image by using an OpenCV vision library, and converting the contour image by using the affine transformation matrix to obtain a template image; calculating pose data of the template image on the conveyor belt, and grabbing a target PCB by using a robot according to the pose data; according to the method, the original image of the target PCB is filtered through adaptive Gaussian filtering and anisotropic diffusion filtering based on whale swarm algorithm control, the PCB edge detection capability of the canny algorithm is improved, and the integrity of the contour features of the PCB is ensured.

The present application belongs to the technical field of situation judgment, and particularly relates to a kind of electrode foil corrosion situation judgment method and system, comprising the following steps: S1, the gray scale image of electrode foil to be analyzed is obtained;Based on the composite gradient reflecting the local brightness and texture information of pixel point, the structure tensor is constructed, and the anisotropic diffusion coefficient is determined according to the structure tensor, the gray scale image is processed by iterative anisotropic diffusion filtering using the anisotropic diffusion coefficient, and the filtered image is obtained;S2, the anisotropic diffusion coefficient determined for each pixel point is fused with the pixel intensity of the filtered image to obtain a high-contrast corrosion significance map.The present application can smooth the complex texture and noise in the background area of electrode foil image while maintaining the outline edge information of the corrosion area, solve the contradiction between denoising and edge preservation in traditional filtering method, make the corrosion situation result more reliable, and improve the automation level of electrode foil product quality detection.

This invention relates to the field of image data processing technology, specifically to a fastener quality traceability method based on the Industrial Internet of Things (IIoT). The method includes: acquiring surface images of multiple fasteners to be inspected and their corresponding timestamps; smoothing the surface images using an improved anisotropic diffusion filtering algorithm to obtain multiple smoothed surface images; performing edge detection on the smoothed surface images to obtain multiple edge contours; obtaining the area of ​​the bounding rectangle of each edge contour; if the area is not within a set allowable range, determining that the fastener to be inspected has a quality defect; and tracing the source based on the timestamps of the surface images of the multiple fasteners to be inspected. This invention solves the problem of low accuracy in quality inspection.

The invention discloses a cardiac death area image detection method and system based on a cardiovascular image, and relates to the technical field of image detection, and the method comprises the steps: obtaining cardiovascular image data; in the current de-noising stage, de-noising processing is carried out on the cardiovascular image data by adopting an anisotropic diffusion filter; evaluating a noise suppression effect according to the denoised cardiovascular image data; according to the noise suppression effect, a de-noising processing strategy is adjusted, and the de-noising processing strategy is used for executing the next de-noising stage; after noise suppression processing of the cardiovascular image data is completed, a denoising quality grade is generated, and the noise suppression processing comprises a plurality of denoising stages; and according to the de-noising quality grade, carrying out myocardial death area image detection on the cardiovascular image data after noise suppression processing. According to the method, fine-grained de-noising processing can be carried out by dividing multiple de-noising stages, the de-noising effect is enhanced, image detection is achieved, and the accuracy and reliability are improved.

The application relates to an anti-NMDAR encephalitis clinical prognosis evaluation method based on artificial intelligence and belongs to the technical field of artificial intelligence and data processing. The method comprises the following steps: acquiring multi-modal neural image data of a patient; adopting a tensor decomposition fusion strategy to perform fusion preprocessing on the multi-modal neural image data, retaining cross-modal spatial correlation through low-rank constraint to obtain an output tensor after fusion; performing lesion-aware anisotropic diffusion filtering on the output tensor after fusion to obtain an output image after diffusion filtering; constructing an anti-NMDAR encephalitis clinical prognosis evaluation model, inputting the output image after diffusion filtering into the model for training, optimizing the training process by adopting an Adam adaptive optimizer, and finally obtaining a trained model; and inputting the output image after diffusion filtering to be evaluated into the trained model to obtain an evaluation classification result. The application can enhance the lesion recognition and classification capability of the model.

The invention provides a step-by-step frequency expanding method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a pre-stack CRP gather; performing interpretive CRP gather superposition on the pre-stack CRP gather to obtain superposed seismic data; performing spectrum equalization processing of a scale domain on the superposed seismic data to obtain a reconstructed seismic image; carrying out smooth filtering processing on the reconstructed seismic image by adopting anisotropic diffusion filtering; selecting a scale range with a high signal-to-noise ratio in the reconstructed seismic image after smooth filtering processing to extrapolate other scale segments to obtain seismic data of a plurality of scale segments; and performing signal reconstruction on the seismic data of all the scale segments to obtain broadband seismic data. According to the method, the relative effective frequency band of seismic data can be remarkably widened, stratum details are well depicted, the resolution of the seismic data is effectively improved, and the method has high industrial practical value and application prospects in oil and gas seismic exploration.

The present application relates to the technical field of image data processing, in particular to a kind of linear guide rail scanning image real-time denoising method and system, comprising: obtaining linear guide rail scanning image to be processed;Using improved anisotropic diffusion filter algorithm to linear guide rail scanning image to be processed is denoised, to obtain the linear guide rail scanning image after denoising processing.Wherein, improved anisotropic diffusion filter algorithm includes diffusion function, the diffusion function is positively correlated with initial diffusion function, the artifact boundary confidence of each pixel point, and the artifact boundary confidence is characterized as the credibility of each pixel point belongs to artifact.The present application solves the problem that existing algorithm is not enough when facing complex noise.

The present application relates to the technical field of fault diagnosis, and more particularly to an air conditioning unit fault monitoring method and system based on multi-modal data fusion. The method comprises: acquiring state values of the air conditioning unit in multiple dimensions in real time; determining the abnormality degree of each dimension at the current time; determining the time step optimization coefficient of each dimension at the current time; determining the adaptive time step of each dimension at the current time; using an anisotropic diffusion filtering algorithm for denoising and obtaining the abnormal score of the air conditioning unit at the current time to realize fault monitoring of the air conditioning unit. The present application dynamically adjusts the time step according to the sorting position of the target dimension state value within the time period of the current time, the median and the coefficient of variation in the ranking, and other indicators, so as to more flexibly and accurately capture the change speed of the state value in different time periods, avoid the defect that the fixed time step in the traditional algorithm cannot adapt to the fast and slow state changes, and make the subsequent fault monitoring more accurate.

The application provides a high-end equipment weld seam radiographic defect semantic reasoning method and system under a few sample conditions. First, U-Net is used to extract the weld seam area and adaptively slide window and cut, and the anisotropic diffusion filtering and the variance guided CLAHE enhancement bottom layer features are combined; a controllable diffusion model of a boundary box perception symbol distance and a category condition is proposed to generate a high-fidelity synthetic defect image. Second, a light-weight visual detection model is used to output defect positioning and classification results. Then, an industrial weld seam knowledge graph is constructed. Finally, a large-small model coordination center of a physical-semantic dual state space is constructed, visual features are mapped to a large language model for semantic review, batch statistics and trend early warning are realized by analyzing the lead type identification, and a customized diagnosis report containing defect type qualitative analysis, mechanism tracing, physical troubleshooting and trend early warning is output. The application realizes a leap from bottom layer visual perception to deep layer causal cognition.

The invention discloses an electric arc multi-dimensional quantitative analysis system and method. The method specifically comprises the following steps: framing an electric arc video and performing time error compensation; taking the arc-free reference image as a benchmark, adopting anisotropic diffusion filtering noise reduction, converting the reference image and the target image into a YUV color space, and generating an arc region mask based on brightness channel difference; a low-temperature area, a medium-temperature area and a high-temperature area are obtained through segmentation in the mask by adopting a maximum between-class variance threshold value; establishing a pixel-level radiation power model, and calculating the energy of each temperature zone and the energy ratio of each temperature zone; extracting a histogram of oriented gradient feature in the temperature partition; recognizing and combining quasi-steady-state intervals by combining time and area double threshold values, a time change rate threshold value and a sliding variance threshold value; and outputting a temperature distribution diagram, a curve of energy and energy ratio along with time and a gradient field thermodynamic diagram. The system is composed of a video framing processing module, an electric arc characteristic acquisition module, a quasi-steady state analysis module and an energy visualization module, and zoning of the electric arc and joint quantification and display of energy and form are achieved according to the process.