752 results about "Image noise" patented technology

The invention relates to the technical field of welding defect identification, and discloses a welding defect identification method and system based on a molten pool image. According to the method, a multispectral high-speed camera is used for collecting a molten pool dynamic image sequence in the welding process, after multi-scale morphological filtering preprocessing is conducted, a fusion feature vector is extracted through a depth separable convolutional network, and then the fusion feature vector is input into a defect classification model adopting a heterogeneous graph neural network architecture to obtain a defect probability distribution matrix. Then, constructing a dynamic sparse optimization model to position defects, generating a defect space coordinate set, and finally, outputting welding defect types and position information through hierarchical verification framework processing. According to the method, the problems of welding image noise interference, complex defect characteristics and the like are effectively solved, the accuracy and reliability of welding defect identification are improved, and powerful technical support is provided for welding quality control.

The invention relates to the technical field of image processing, in particular to a road area intelligent extraction system for unmanned aerial vehicle aerial images, and the system comprises an image noise filtering module which analyzes and inputs frequency domain information of the unmanned aerial vehicle aerial images, recognizes a target noise mode, and filters the unmanned aerial vehicle aerial images. According to the method, frequency domain noise analysis and notch filtering are combined, the sharpness of the road edge is reserved, meanwhile, high-frequency noise and periodic interference are restrained, and the visibility of a low-contrast road area is improved; local pixel spectrum dispersion calculation is introduced and is dynamically fused with morphological parameters, and direction consistency and length-width ratio constraints are combined, so that the connectivity perception capability of the fragmented non-paved road is enhanced, and the fracture misjudgment under the form-changeable terrain is reduced; a normalized vegetation index and multi-band reflectivity are combined for modeling, water reflection interference is eliminated through a blue light band standard deviation, and a space mask of a non-road surface feature interference area is constructed.

The invention relates to a financial bill management system and an implementation method thereof. The financial bill management system comprises a data acquisition module, an AI identification module, an auditing module, a cost reimbursement module, a report generation module and a data storage and backup module. According to the financial bill management system and the implementation method thereof, by introducing the AI recognition module, automatic recognition and classification of bill types are achieved, the requirement for manual operation is reduced, the working efficiency and accuracy are improved, the AI recognition module can automatically process bill information, the whole process from image preprocessing to final key information extraction is highly automatic, and the efficiency is high. The system is not limited to bill identification and classification, but also covers a plurality of functional modules such as data acquisition, auditing, expense reimbursement processing, report generation, data storage and backup and the like, and provides a one-stop solution for users. The image preprocessing unit can effectively remove image noise, correct inclined images, cut redundant parts and enhance image contrast and brightness.

The invention discloses a federated learning model training method and system based on differential privacy image noise addition. The method comprises the following steps: S1, initializing an evaluation network, a target network and hyper-parameters; s2, initializing a global model and parameters; s3, acquiring a selected client and a privacy budget; s4, the selected client performs local training and uploads the disturbance gradient to the central server; s5, the central server aggregates the disturbance gradient of the client to obtain an updated global model and a reward corresponding to the current state; s6, the next state of federated learning is obtained, and the evaluation network and the target network are updated; s7, repeatedly executing the steps S3 to S6 until the set number of global iterations is reached; and S8, updating the exploration rate and repeatedly executing the steps S2 to S7 until the set federal learning times are reached. The invention provides a dynamic differential privacy method in combination with a federal learning framework, which can improve the image recognition rate while ensuring the privacy of the user.

The invention provides an image noise mark feature selection method and system, a storage medium and a computer. The method comprises the steps of obtaining a to-be-processed image noise mark data set; embedding a sample set in the image noise mark data set into a multi-granularity fuzzy cluster to construct a dynamic fuzzy membership evaluation matrix; dynamically evolving a multi-level high-precision granular ball cluster; obtaining mark distribution with high identification degree; constructing a rough perception feature evaluation framework based on granular ball topology driving, extracting decision equivalence classes by combining rough set upper and lower approximation and extended positive domain models, and determining and measuring the contribution degree of each feature to a decision system by fusing multi-granular-ball decision boundary information based on a dependency degree quantitative model; a particle and ball structure consistency verification mechanism is introduced, and multi-level evaluation is carried out on the importance of the features through dependency and consistency. According to the method, the optimal feature subset with strong anti-noise performance and high discrimination capability is obtained, and stable and efficient input support is provided for a subsequent image noise mark learning model.

The invention relates to the technical field of pattern recognition, in particular to an ecological restoration nursery stock quality detection system based on pattern recognition, which comprises an image acquisition module for controlling camera equipment to acquire nursery stock images under target time and illumination conditions to obtain an original image set; and performing format unification and cleaning on the original image set, removing image noise and irrelevant information, and generating a cleaned image set. According to the method, the image pickup equipment is controlled to obtain the original nursery stock image under the target time and illumination condition, the consistency and stability of image data are improved, after unified format and information cleaning are carried out on the original image, image noise and invalid information are eliminated, and the nursery stock feature extraction precision is improved. Based on multi-dimensional feature extraction of colors, textures and shapes and comparative analysis of known quality standards, accurate judgment of seedling growth quality is realized, and scientificity and reliability of seedling screening are improved.

The invention provides a dolomite multi-scale digital core reconstruction method and device, electronic equipment and a medium, and the method comprises the steps: inputting a to-be-processed CT scanning two-dimensional slice image of a dolomite sample into a reconstruction model, and obtaining a multi-view SEM style image and view parameters; inputting the style image and the visual angle parameter into a NeRF model to obtain a three-dimensional high-resolution image; the reconstruction model is obtained based on the following steps: adjusting a CT scanning two-dimensional slice image and an SEM image of a dolomite sample according to a unified standard; inputting a CT scanning two-dimensional slice image and an SEM image which are in a unified standard into a FastGAN network for preprocessing of image enhancement and augmentation to obtain a standard simulation image; and training a generator in the CycleGAN network based on the standard simulation image to obtain a reconstruction model. The method can solve the problem that a traditional digital core technology is limited by image noise and equipment cost, and nanometer and micrometer multi-scale fusion is difficult to achieve.

The invention discloses an infrared ship wake detection method and system combining deformable convolution and an attention mechanism. According to the method, the wake detection model based on yov8 improvement is provided, the image noise problem and the small target detection problem under the complex sea condition are comprehensively considered, the deformable convolution, the attention mechanism and the multi-scale feature fusion module are provided, the model detection performance is improved, and therefore the infrared ship wake detection precision is improved. The method is mainly characterized in that a DCNv3-based improved feature extraction module is provided for solving the problems that the shape of a ship wake target is irregular, marine image noise is caused and a small target is difficult to detect, and an improved MLCA attention mechanism is used for improving the attention degree of a model on the target; and an improved multi-scale feature fusion module GFPN is used to improve the detection precision of the small target of the model.

The invention discloses a brain lesion nuclear magnetic image Transform diffusion segmentation method and a brain lesion nuclear magnetic image Transform diffusion segmentation system based on symmetry prior. The invention provides a novel prior-guided diffusion denoising brain lesion segmentation Transform model, long-distance coding features are extracted from an image, local information related to medical image prior features is supplemented, and frequency domain screening fusion coding components are further introduced. A U-shaped denoising model is designed based on a transform network structure, and in order to maintain accurate lesion-related semantic details, a prior feature coding branch is introduced. A priori feature coding branch considers modal-related voxel distribution characteristics and spatial symmetry in a modal, so that a priori guidance attention mask is learned, and attention weights of a pixel level and a patch level are dynamically adjusted. In order to eliminate image noise and a low contrast effect, in a decoding part, a three-branch wavelet frequency spectrum mixed attention module is designed for fusing and screening frequency spectrum feature embedding, and a stable segmentation result is further generated.

An image processing method, performed by a computer device, comprising: extracting sketch texture features at multiple scales from a sketch image; extracting image noise features at multiple scales from preset noise; determining color guide information corresponding to the sketch image; encoding, for each scale, a noise feature based on a sketch texture feature and the color guide information to obtain multi-scale image features; and performing multi-scale decoding on these image features to obtain a colored image comprising a sketch texture corresponding to the sketch image and a color based on the color guide information. A related training method and apparatus are also provided to develop models for this image processing technique.

The invention relates to the field of computer vision, in particular to an image feature recognition method based on computer vision, which comprises the following steps of: preprocessing an input image, removing image noise and correcting image gray deviation to obtain a preprocessed image; performing multi-scale feature extraction on the preprocessed image, obtaining image local texture features and global contour features under different scales, performing dynamic weight fusion, calculating dynamic fusion weight based on a feature response value and a local complexity factor, and performing dimension reduction processing on fusion features through a principal component analysis algorithm to obtain a target feature set; and matching the reference features, calculating the Euclidean distance between the target feature set and the reference features, determining an adaptive matching threshold, judging a matching result, and completing image feature recognition. According to the method, through adaptive filtering and gray level equalization preprocessing, Gaussian pyramid multi-scale feature extraction and dynamic fusion and adaptive threshold matching, feature extraction accuracy, characterization capability and recognition robustness are improved.

The invention discloses an intelligent scalp detection system and a detection method thereof, relates to the technical field of intelligent scalp detection, and aims at obtaining shooting conditions and definition evaluation through real-time environment perception and quality detection, solving hair occlusion and image noise by using deep learning segmentation and directional enhancement, showing a scalp deep structure by means of multi-modal fusion and splicing, and improving the detection accuracy. A diagnosis and optimization strategy is output in comprehensive analysis and self-adaptive closed-loop feedback; through software algorithm level expansion, online training and multi-label pathological expansion can be performed on multi-user long-period data, and adaptive evolution and group statistical analysis are realized; according to the method, the image acquisition stability and segmentation accuracy can be effectively improved, the adaptability of detection to complex illumination and multi-band information is enhanced, the risk of pathological misjudgment such as grease, inflammation and fungi is reduced, and personalized nursing and wide health management application are supported.

The invention relates to an object orientation detection method and device, computer equipment and a storage medium, which are used for solving the problem of poor accuracy and reliability of object pose adjustment in an existing industrial assembly line. The method comprises the following steps: carrying out image acquisition on a target object through a super-resolution full-spectrum imaging device to obtain a plurality of original image sets; carrying out image noise removal, enhancement and edge detection on the image set to obtain a preprocessed image set; inputting the preprocessed image set into a sparse self-encoding convolutional neural network for feature extraction to obtain object orientation feature data; identifying key feature points of the target object by using improved generalized Hough transform; and estimating the orientation of the object based on a fuzzy logic orientation discrimination algorithm, and generating an operation strategy. Through deep learning feature extraction and fuzzy logic orientation discrimination, accurate detection of the object orientation in a complex industrial environment can be realized, and the automation level and stability of assembly line work are improved.

The invention relates to the technical field of machine vision and product defect visual prediction, in particular to a product defect visual prediction method and device based on X rays. According to the method, edge extraction and entropy-based analysis are utilized to enhance defect positioning, meanwhile, the challenges of image noise and non-uniform intensity change are solved, and in order to improve robustness, a self-adaptive threshold strategy combined with DBSCAN clustering is adopted to distinguish defects and noise. Therefore, the problems that in the prior art, the ratio of product defects is low, the visual prediction efficiency of edge blurring is low, and the accuracy is not high are solved. Compared with the prior art, noise reduction, edge detection and crack identification are enhanced. By integrating X-ray image processing, the method improves accuracy, reduces false positive and improves detection efficiency.

The invention discloses a large space positioning fast conversion method based on image recognition, and relates to the technical field of large space positioning, and the method comprises the following steps: large space image collection and deployment, on-site measurement of the size of a large space and the position of an obstacle, introduction of software to construct a basic three-dimensional model, simulation of an initial site of a camera, and calculation of the initial site of the camera. Determining a final layout through evaluation and screening, and performing installation and debugging; according to the invention, through comprehensive utilization of an image identification technology and a multi-camera cooperative positioning mechanism, precision and reliability of target positioning in a large space are substantially improved; through combination of a triangulation principle and a Kalman filtering technology, the position of a target in a three-dimensional space can be effectively deduced reversely, a positioning result is corrected in real time, and positioning jitter caused by image noise and transient shielding of the target is smoothed; the positioning continuity is improved, and the stability and accuracy in a complex environment are enhanced.

The invention relates to the field of tea disease detection, in particular to a tea disease detection method based on a lightweight YOLOv8 model. According to the technical scheme, the method comprises the steps of collecting various types of tea disease images; the method comprises the following steps: preprocessing multiple types of collected tea disease images, including illumination normalization, geometric enhancement, illumination enhancement, adaptive histogram equalization, a multi-scale Retinex algorithm, dynamic Gamma correction and image noise removal, and marking scab regions in different types of tea disease images to obtain a training data set; a lightweight YOLOv8 model is constructed; using the training data set to train the lightweight YOLOv8 model; and inputting a collected tea disease image into the trained lightweight YOLOv8 model, and outputting a tea disease type through the trained lightweight YOLOv8 model. The method is suitable for tea disease detection.

The invention discloses a device and a method for detecting the surface flatness of a wood-plastic plate, and the method comprises the steps: carrying out the image analysis of a surface image of the wood-plastic plate through employing a computer vision technology based on deep learning, removing the image noise through gray conversion and Gaussian filtering, extracting stable image texture distribution feature representation, and meanwhile, carrying out the detection of the surface flatness of the wood-plastic plate. The embedded representation of the statistical parameters of the image texture features is used as auxiliary information, and the multi-scale texture information fusion representation of the image texture distribution features and the texture statistical features is obtained through cross-domain joint coding, so that intelligent detection of the flatness of the wood-plastic plate is carried out. Through the mode, the anti-interference capability of detection can be effectively improved, and the recognition of illumination variation and small defects on the surface of the plate is enhanced, so that the detection precision and reliability are improved, and the requirements of modern industrial production are met.

The invention discloses a processing method for capturing an object image in real time based on a computer vision technology, and relates to the technical field of computer vision, and the method comprises the following steps: 1, capturing an object image in real time through a USB camera and a network camera, and transmitting image data to a server side in real time through TCP / IP for processing; step 2, removing noise by using a Gaussian filter, enhancing contrast by histogram equalization, extracting image features through a hybrid architecture based on Inception V3 and ResNet-50, optimizing feature extraction through a multi-scale attention mechanism by the hybrid architecture, performing pixel-level segmentation by using a full convolutional network, dividing the image into different object areas, and performing image segmentation on the object areas; post-processing is carried out on the segmentation result; according to the method, the Gaussian filter and the histogram equalization technology are adopted, image noise is effectively removed, the contrast ratio is enhanced, image features are extracted through the Inception network, pixel-level segmentation is carried out in combination with the full convolutional network, and high-precision and high-efficiency object segmentation is achieved.

The invention discloses a binocular camera global ultra-sensitive depth direction vibration measurement method, and relates to the technical field of structural vibration measurement. The method comprises the following steps: completing internal reference and external reference calibration of a binocular camera based on a Zhang's calibration algorithm, and ensuring the three-dimensional reconstruction precision; in image processing, effective pixels with high dynamic response are screened by calculating pixel intensity time-history variance, a pixel intensity space-time matrix is constructed, and singular value decomposition is adopted to extract front k-order modals to reconstruct the pixel intensity matrix so as to reveal real signals originally submerged in imaging noise and rounding errors. And multi-direction cost aggregation is carried out by using a semi-global matching algorithm, the parallax calculation robustness in a low-contrast scene is improved, and a high-precision depth map is generated. And depth information is converted into three-dimensional space information in combination with camera parameters, so that non-contact measurement of the vibration time travel curve in the full-field depth direction is realized.

The invention provides a contour extraction method for a graph in a scanning electron microscope image and a related product. The method comprises the following steps: acquiring a scanning electron microscope image of a to-be-extracted contour and a design layout corresponding to the scanning electron microscope image; carrying out contour extraction on a graph in the scanning electron microscope image to obtain an initial graph contour; according to the initial graph contour, a design graph on the design layout is zoomed out or zoomed in, and a zoomed design graph is obtained; and performing contour extraction on a graph in the scanning electron microscope image by using the scaled design graph to obtain a target graph contour. After the design graph is scaled according to the proportion, the size of the design graph is quite close to the size of the actual graph. On the basis, contour extraction of the scanning electron microscope image is carried out, the success rate and the precision of graph contour extraction can be improved, and wrong contour extraction caused by image noise and signal intensity is avoided.

The invention discloses a complex environment image semantic segmentation method based on a fine-grained enhancement network, and the method comprises the steps: obtaining a to-be-detected environment image, inputting the to-be-detected environment image into a pre-trained fine-grained enhancement network model, and obtaining a semantic segmentation result of the to-be-detected environment image. According to the method, a fine-grained enhancement network model constructed through optimization is utilized, global feature information and local feature information are subjected to deep extraction and multi-scale fusion through a backbone network layer, the feature extraction capability of the model is enhanced, and then the feature extraction capability of the model is enhanced by means of up-sampling and detail enhancement processing of a neck network layer. The detail recovery capability of the model is enhanced, the image noise is effectively reduced, the semantic segmentation result of the environment image is obtained through the detection head, and accurate semantic segmentation for the complex environment image is realized.

The invention relates to the technical field of material analysis, in particular to a method for evaluating a pore structure of a refractory material by using industrial CT, which comprises the following steps: selecting an X-ray source parameter, a detector parameter, a focal length and the like of the industrial CT to be matched with a refractory material with different thicknesses, selecting a filter to avoid image noise, identifying an internal structure of the refractory material, and evaluating the pore structure of the refractory material. And obtaining a two-dimensional projection image of the internal structure. According to the method, the image noise is reduced through the filtering technology, the internal structure of the refractory material can be identified more accurately, the data integrity is ensured, a more accurate three-dimensional model can be generated, the analysis of the pore structure is more detailed and accurate, and by quantitatively analyzing the pore size, shape and distribution of the refractory material, the analysis accuracy of the refractory material is improved. The performance change of the material can be effectively monitored, the prediction capability of the refractory material in practical application is improved, and the application safety and reliability of the refractory material in an extreme environment are enhanced.

The invention relates to a circuit board defect identification method and system based on multi-dimensional image data, and belongs to the technical field of data identification processing, and the method comprises the following steps: obtaining synchronous image data of a circuit board to be detected in a plurality of imaging modes; performing space-spectrum joint registration on each modal image to generate a multi-dimensional image cube with a unified coordinate system and pixel alignment; inputting the multi-dimensional image cube into a pre-trained multi-branch heterogeneous fusion neural network; generating a pixel-level defect probability graph by utilizing a defect sensing context decoder, and performing geometric constraint optimization on the probability graph by combining prior information of a circuit board design layout; outputting defect types, positions and confidence coefficients, and establishing an interpretable defect fingerprint database according to the multi-dimensional response characteristics of the defects; the method has the beneficial effects that false defect signals generated by image noise and circuit board surface texture interference can be effectively inhibited, the omission ratio and the false detection ratio are greatly reduced, and pixel-level accurate defect positioning is realized.

A training data generation method that is performed by a computing device including at least one processor according to an embodiment of the present disclosure includes: generating random noise corresponding to the noise of an input image; generating first noise and second noise based on the random noise; and generating a first image based on the input image and the first noise, and generating a second image based on the input image and the second noise; and the first and second images are noise-independent of each other.

The invention discloses an infrared and visible light image fusion method for an inspection robot. The infrared and visible light image fusion method specifically comprises the following steps: S1, pre-training an infrared image noise suppression network NANet and a visible light image light intensity sensing network LIPNet; s2, using an IF Block and an EF Block based on Transform to extract image features, and using the IF Block and the EF Block based on Transform to extract image features; s3, calculating an adaptive weight map by using an adaptive weight estimation module (AWEM), and guiding feature retention by using the weight map; and S4, realizing feature reconstruction by using a feature fusion module formed by three convolutional layers, and generating a fused image. According to the infrared and visible light image fusion method for the inspection robot disclosed by the invention, the performance of the fused image in an advanced visual task is effectively improved, and the influence of noise and light pollution on the fusion performance is effectively reduced by performing comparative analysis on a public data set and various mainstream fusion methods from subjective and objective angles; and the salient target and background texture information can still be fully reserved in a severe scene.

The invention particularly relates to a liquid level monitoring system for a flow battery based on machine vision. The liquid level monitoring system comprises an image acquisition module; the liquid level detection module is used for analyzing based on the liquid level image of the flow battery to obtain an analyzed and evaluated liquid level of the flow battery; the display and interaction module is used for outputting and presenting the liquid level detection result; and a system control module. According to the method, the first evaluation liquid level preliminarily locates the liquid level contour in a rapid gradient calculation mode, the second evaluation liquid level accurately extracts the edge through multi-stage processing, and the two parts complement each other; in the comprehensive analysis stage, by means of unique mean value calculation, a method for constructing a geometric model based on liquid level deviation and calculating a difference coefficient, and in combination with a preset threshold value, an additional liquid level value is matched, errors of a single detection method can be effectively eliminated, and detection deviation caused by factors such as environment interference and image noise can be corrected.

The invention relates to the field of hearth flame detection, and particularly discloses a hearth flame detection device and method, and the method comprises the following steps: a, capturing a flame image in a hearth through a high-resolution infrared thermal imager, and transmitting the flame image to an image processing unit; step b, executing the following operations in the image processing unit to evaluate and optimize the image quality: removing noise in the image by applying an adaptive median filtering algorithm, wherein the algorithm effect is quantified by calculating a flame image noise removal effect evaluation value Enoise (l); the flame state parameters are monitored in real time, once the parameters are lower than the preset threshold value, the safety pre-warning system is triggered immediately, the real-time monitoring and pre-warning mechanism helps to find and handle potential problems such as instability or extinguishment of flames in time, and therefore safety accidents are effectively prevented.

The invention discloses a multi-agent PCB (printed circuit board) design document automatic extraction and table completion method. According to the method, the table area in the PCB design document is automatically extracted by training the YOLOv11 model, and missing table lines are complemented by using an image processing algorithm, so that the extraction precision of table data is improved. A multi-agent architecture is adopted, text extraction and table data extraction and combination are carried out respectively, and a clear document is finally generated and output. According to the method, the defects of an existing OCR technology in table extraction can be overcome, especially the accuracy problem under the influence of table line missing or image noise can be solved, and the method has high practical value and innovativeness.

The invention provides a method for repairing an underwater degraded image. The method comprises the following steps: S1, inputting the underwater degraded image; s2, constructing an unsupervised deep learning image reconstruction network based on a physical model; s3, constructing a reconstruction network consistency verification module according to an underwater image imaging physical model and the network output reconstruction degraded image; s4, according to an image equivariant theory, constructing a fidelity image network equivariant loss and an underwater imaging equivariant loss; s5, according to the image isomorphism, constructing a fidelity image to restore an image isomorphism verification module; s6, according to the noise distribution of the underwater degraded image, constructing noise loss by using an unbiased estimation method, and correcting the deviation of reconstruction loss; s7, according to a gray world hypothesis, compensating the information of the blue channel and the green channel with relatively small color weakness to the red channel, and constructing a color fidelity verification module; and S8, carrying out constraint training on the network according to the steps S3-S7 until loss convergence, and obtaining a fidelity image estimation module for repairing the degraded image. The invention further provides a storage medium and electronic equipment.

The invention belongs to the technical field of medical image processing, and particularly relates to a medical image processing method and system based on multi-modal fusion, and the method comprises the steps: obtaining a CT image and a corresponding digital pathological image, and carrying out the pairing of the CT image and the corresponding digital pathological image; respectively inputting the paired CT image and digital pathological image into a CT branch and a pathological branch; the CT branches extract macroscopic morphology and internal density distribution to obtain a CT feature map, and the pathological branches extract microscopic details to obtain a pathological feature map; the CT feature maps and the pathological feature maps from different layers are fused in a layer-by-layer cascade and feature splicing mode, and a comprehensive feature map is obtained. The method can still stably extract and identify key image features and modes in the face of image noise, different imaging devices or image visual differences among individual samples, and has high image analysis robustness.