206 results about "Key images" patented technology

The invention discloses a valve state monitoring and safety interlocking method based on machine vision, and belongs to the technical field of industrial process control and safety. The method comprises the following steps: 1, calibrating a camera and establishing a model; step 2, image acquisition and preprocessing; step 3, key image feature extraction and state decision; fourthly, opening degree calculation and state judgment are carried out; 5, safety interlocking is triggered; and 6, feeding back data. The valve state monitoring and safety interlocking method is achieved through the valve state monitoring and safety interlocking system. According to the invention, the millimeter-level or even submillimeter-level displacement and angle real-time accurate measurement of the opening degree of the valve can be realized, the state monitoring of the valve is realized, and the stability and reliability in the production process are improved; state judgment is carried out by comparing the real-time opening value with the instruction opening from the control system, a safety interlocking signal is generated and transmitted to the safety interlocking system, and finally physical safety action is triggered, so that a set of complete safety protection system is realized, and accidents are avoided.

The invention provides a DICOM (Digital Imaging and Communications in Medicine) file sequence processing method, a medical image analysis method, system and device, a computer storage medium and a computer program product, the processing method comprises the following steps: loading a DICOM file sequence corresponding to a series of image sets obtained by tomography, and arranging according to a scanning sequence to obtain an initial image file queue; reading DICOM files from a specified position to form a key image file queue with a fixed length; and reading sequence-level DICOM metadata and image-level DICOM metadata for representing key features from the key image file queue, and arranging the sequence-level DICOM metadata and the image-level DICOM metadata according to a preset rule to obtain a sequence DICOM metadata queue with fixed length and fixed information items of each position. By adopting the scheme provided by the invention, the obtained metadata queue with the fixed dimension provides a data basis for subsequent deep learning network input, and improvement of the adaptability with the deep learning network and improvement of the analysis accuracy are facilitated.

The invention relates to an image processing method, device, equipment and system for a double-layer metasurface. The method comprises the following steps: during decryption, using a bottom-layer metasurface unit in a double-layer metasurface and adopting electromagnetic waves of a first frequency to respectively restore a plurality of first password book images acquired from different terminals to obtain a plurality of second password book images; using the bottom layer metasurface unit and the covering layer metasurface unit in the double-layer metasurface and adopting electromagnetic waves of a second frequency to restore the obtained first key image to obtain a second key image; and determining a second original image according to the plurality of second password book images and the second key image. According to the technical scheme provided by the invention, the second password book image and the second key image are respectively determined by adopting two different frequencies. The cracking difficulty is increased, and the safety is improved.

The invention belongs to the technical field of underwater image processing and crack detection, and particularly relates to a hydropower station dam underwater crack image positioning method, which comprises the following steps of: acquiring image data of a dam surface crack and acquiring a key image; the method comprises the following steps: preprocessing an acquired underwater image, detecting a crack region of the preprocessed image by adopting a crack target detection network based on YOLOv9, and outputting boundary frame coordinates and confidence of a crack; a post-processing algorithm is used to optimize a detection result, and classification is carried out according to fracture morphological characteristics after multi-frame fusion; storing crack detection and positioning data to a database, and displaying the spatial distribution and change trend of cracks; and establishing an early warning model based on the crack propagation rate and the morphological change, and if early warning is triggered, generating a risk assessment report. The method overcomes the defects that traditional manual inspection is low in efficiency, is greatly interfered by water quality and is difficult to cover hidden areas, and the automation level and reliability of health monitoring of the dam structure are remarkably improved.

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.

A medical image processing apparatus includes: an image acquisition unit that acquires a plurality of medical images; an image classification unit that classifies the medical images into at least one of a plurality of categories; an image display unit that displays, on a screen, at least one of the plurality of medical images based on a result of the classification as an automatically selected image; an input receiving unit that receives an input from a user to select an image requiring reselection as a key image from among the displayed automatically selected images as a selected image; and a display control unit that switches and displays, as a switching image, a non-automatically selected image other than the automatically selected image among the plurality of medical images classified into each category, instead of displaying the selected image.

The invention provides a semantic communication method and device based on a generative AI large model, equipment and a storage medium, and relates to the technical field of communication. Semantic segmentation is carried out on a to-be-transmitted image to obtain a key image area and a non-key image area, an undistorted transmission strategy is adopted for the key image area, and a text generation model is used for the non-key image area to generate a concise description text; and the receiving end reconstructs the non-key image region according to the description text of the non-key image region, and generates a reconstructed image comprising the original pixel data of the key image region and the synthetic pixel data of the non-key image region. The technical problem that in the prior art, accuracy and reliability of semantic communication are limited is solved, bandwidth resources are relieved, transmission efficiency is improved, and efficient transmission with high fidelity of the key image area and complete semantics of the non-key image area is achieved.

The invention relates to the technical field of image monitoring, in particular to an intelligent visual monitoring method and system in an extremely cold environment, and the method comprises the following steps: based on an extremely cold monitoring environment, collecting environment temperature data through a temperature sensor, measuring the wind speed and the snowfall density, and obtaining low-temperature environment state data; according to the invention, environment state data is acquired in real time, a battery charging and discharging strategy is automatically adjusted to adapt to low-temperature influence, a heating element is automatically adjusted to prevent the equipment from frosting and freezing, normal operation of functions of the monitoring equipment is ensured, and light sensitivity and exposure parameters of an image sensor are adjusted, so that the monitoring efficiency is improved. The image quality is improved, the influence of an extremely cold environment on the image quality is eliminated, the monitored data is more accurate and reliable in vision, the compression ratio is dynamically adjusted, key image data is preferentially transmitted, and the efficiency and the real-time performance of data transmission are enhanced.

Disclosed herein are methods and systems for using artificial intelligence modeling techniques to generate a path for an ego. In an embodiment, a method comprises retrieving image data of a space around an ego, the image data captured by a camera of the ego; predicting by executing an artificial intelligence model, an occupancy attribute of a plurality of voxels corresponding to the space around the ego; generating a 3D model corresponding to the space around the ego and each voxel's occupancy attribute; upon receiving a destination, localizing, by the processor, the ego by identifying a current location of the ego using a key image feature within the image data corresponding to the 3D model without receiving a location of the ego from a location tracking sensor; and generating a path for the ego to travel from the current location to the destination.

The invention discloses a nasopharyngeal carcinoma necrosis prediction method and device based on feature fusion and a medium, and the method comprises the steps: obtaining multi-modal medical data of a target patient at a plurality of preset time points, and extracting medical features in different modals for each time point, so as to form a medical time sequence feature matrix in different modals; inputting the medical time sequence feature matrixes in the different modes into a gated twin-tower Transform network, and performing feature fusion on the medical time sequence feature matrixes in the different modes through a preset inter-channel-time step cross attention mechanism to obtain key image fusion features and nasopharyngeal carcinoma necrosis risk prediction results; and if the nasopharyngeal carcinoma necrosis risk prediction result is greater than a preset risk threshold, inputting the key image fusion features and the original three-dimensional image data into a 3D nnU-Net network to generate a nasopharyngeal carcinoma necrosis prediction map of the target patient. The prediction accuracy of nasopharyngeal carcinoma necrosis can be improved.

The invention relates to the technical field of remote sensing image processing, in particular to a homomorphic filtering-CLAHE remote sensing image enhancement method based on integration strategy multi-target particle swarm optimization, which comprises the following steps: acquiring remote sensing image data, and preprocessing the remote sensing image data; converting the preprocessed remote sensing image into an HSV color space, and extracting a V component in the HSV color space; constructing a frequency domain-spatial domain hybrid enhancement framework of homomorphic filtering and contrast-limited adaptive histogram equalization, and performing enhancement processing on a V-channel component by using the framework; constructing an integrated strategy multi-target particle swarm optimization algorithm; constructing a four-target fitness function including structural similarity, average gradient, information entropy and gray variance, and guiding particles to search in the direction of optimizing a plurality of key image quality indexes at the same time by using the function so as to realize optimal selection of remote sensing image enhancement parameters; the method can effectively enhance the definition and structural integrity of the terrain texture in the remote sensing image of the complex mountainous area.

The invention relates to the technical field of fire science and computer vision, in particular to a long-term fire heat release rate prediction method based on fire scene image recognition. The method comprises the following steps: selecting a plurality of groups of fire video data from an NIST fire calorimetric database, preprocessing the fire video data, and extracting key image frames from image frames corresponding to the selected fire video data by adopting a dense frame extraction mode, the HRR prediction method comprises the following steps: extracting a plurality of key image frames, marking the heat release rate of the corresponding key image frames, establishing a deep learning model based on Bi-LSTM and Attention mechanisms, inputting the plurality of extracted key image frames into the deep learning model, extracting spatial-temporal characteristics by the deep learning model, and generating an HRR prediction value corresponding to a future fire image. According to the method, the time sequence features and the spatial features of the fire scene image sequence are extracted, and the attention mechanism and the backbone architecture of the bidirectional long-short-term memory network are combined, so that the high-precision prediction of the long-time history and the long-time-history heat release rate of the fire can be realized.

The invention relates to the technical field of satellite communication, and discloses an efficient encoding and decoding and distribution transmission method for satellite image data, which comprises the following steps of: acquiring satellite epoch parameters, estimating a downlink channel capacity sequence, decomposing the image data into a basic feature code stream and a background residual code stream, and distributing a transmission priority, and mapping the basic feature code stream to the most significant bit of the modulation symbol, mapping the background residual code stream to the least significant bit of the modulation symbol, and implementing nonlinear displacement stretching on the coordinate of the constellation point where the modulation symbol is located according to the carrier frequency shift change rate. A core semantic bit phase tolerance boundary is expanded by compressing a background data judgment space, high-fidelity transmission of key image features in a severe phase distortion environment is ensured, and the information transparent transmission capacity in a high-dynamic link is improved.

The invention discloses a quality evaluation method and system for an optical lens of a display device, and the method comprises the steps: obtaining and preprocessing an output image of the optical lens, analyzing the output image, determining key imaging features which affect the quality, and determining the feature value of each key imaging feature; evaluating the quality of the optical lens based on each key imaging feature and the feature value to obtain an initial quality evaluation value; determining an influence coefficient of each key imaging feature on the imaging quality of the optical lens, and determining a comprehensive influence coefficient based on the influence coefficient; and adjusting the initial quality evaluation value of the optical lens based on the comprehensive influence coefficient to obtain a final quality evaluation value of the optical lens. According to the method, the imaging characteristics output by the optical lens are quantitatively evaluated, and the evaluation is adjusted according to the influence of the imaging characteristics on the imaging quality of the optical lens, so that the imaging quality of the optical lens can be comprehensively and accurately evaluated, the objectivity and efficiency of evaluation are improved, and a scientific basis is provided for quality management and performance optimization.

The application discloses a financial document multi-agent collaborative analysis method and device, a medium and an electronic equipment, and relates to the technical field of financial technology. The method comprises the following steps: preprocessing a financial document to be analyzed to obtain preprocessed image data; detecting the preprocessed image data by using a pre-trained target detection model to obtain a key image region; and performing multi-round collaborative analysis on the key image region by using multiple image recognition agents and conflict decision agents to obtain a collaborative analysis result. The method can improve the accuracy of financial document data analysis and obtain logically consistent and highly reliable structured data.

The embodiments of the present disclosure provide a visual positioning method and apparatus, a computer-readable storage medium, and an electronic device for indoor navigation, and belong to the field of computer and communication technology. The method comprises: obtaining an indoor query image; obtaining feature points of the query image; obtaining a descriptor of the feature points of the query image; querying an indoor map based on the descriptor of the query image to obtain a set of candidate key images in the indoor map of the query image; comparing each feature point of the query image with the feature points in each key image in the set of candidate key images, and filtering the feature points of the query image at optimal and suboptimal ratios to obtain the feature points of the filtered query image; and determining the position and posture when the query image was taken based on the feature points of the filtered query image. The method of the present disclosure can achieve visual positioning indoors.

The invention discloses a three-dimensional cultural relic 3D scanning ultra-high-definition enhancement method, device and equipment and a medium, and relates to the technical field of three-dimensional modeling, and the method comprises the steps: obtaining a light supplement enhancement image set, carrying out the dense matching strategy of the light supplement enhancement image set, and generating image pairs with consistent structures; minimum coverage rectangular region extraction is carried out on the image pairs with the consistent structures, and redundant complementation region pairs are generated; performing texture completion processing on the receptor region image in the redundant completion region pair to obtain a texture completion image set; and key image pairs are extracted based on the final texture completion image set, a matching point pair set is obtained through scale invariant feature transformation, and a matching point three-dimensional coordinate set is obtained through triangulation of each matching point pair. According to the invention, pixel-by-pixel brightness compensation and color equalization are carried out by combining the time-weighted exposure consistency index, so that illumination equalization adjustment of the region with uneven local exposure is realized.

The application discloses a lightning target recognition model and method facing image input modal imbalance and high false alarm rate, and belongs to the technical field of image recognition and meteorological disaster monitoring. Firstly, Gaussian mixture density estimation and time-weighted fusion are performed on ground-based lightning location data to convert the image label with spatial probability distribution characteristics; then, Himawari-8 satellite multi-channel brightness temperature images and their derived feature maps and radar echo images are used as multi-source heterogeneous inputs. In view of the information imbalance between different observation modalities, an MDE-UNet deep learning model is constructed, a multi-scale feature fusion module, a MLP enhanced decoding unit based on a weighted sliding window and a radar image information enhancement module are innovatively designed, the key image features are enhanced and irrelevant noise is suppressed, and the feature imbalance problem in heterogeneous data fusion is solved. The model adopts an asymmetric weighted BCE-DICE loss function to strengthen the attention to lightning target pixels.

The invention relates to controller power-on test equipment, a test method and a manufacturing system. The controller power-on test equipment comprises one or more power-on test parts, a multi-axis displacement unit, an image acquisition part and an image test part, the power-on test part is used for placing, positioning and fixing a target test controller and powering on the target test controller through the test interface; the multi-axis displacement unit is used for executing multi-degree-of-freedom motion according to a preset or dynamically adjusted acquisition track; and the image acquisition part is used for acquiring test image data displayed by one or more image test parts according to preset shooting parameters in the motion process of the multi-axis displacement unit and transmitting the test image data to a target terminal in real time. According to the method and the device, the acquisition track can be dynamically adjusted according to the test image data, and the test stage of the target test controller is identified based on the continuous image feature change, so that key image acquisition and test judgment are completed in a proper stage, automation and adaptive detection of the power-on test process are realized, and the power-on test effect is improved.

The application discloses a nasopharyngeal carcinoma prognosis feature determination method, system and device and a storage medium. The method comprises the following steps: pathological image preprocessing, color normalization based on dye separation is used to standardize the dyeing of the pathological image; a segmentation network is used to automatically segment the lesion area of the preprocessed pathological image to obtain a segmented image; the segmented image is cropped to obtain a target image block; principal component analysis is used to reduce the dimension of the target image block to obtain reduced dimension data; a clustering algorithm is used to perform unsupervised autonomous learning on the reduced dimension data to obtain pathological image features; and finally, the pathological image features are screened through feature inspection to determine a prognosis pathological feature set. The application can obtain and screen key image features of pathological images closely related to local area recurrence and distant metastasis of nasopharyngeal carcinoma from pathological images to assist in prognosis prediction of nasopharyngeal carcinoma, and can be widely applied to the technical field of image processing.

The application discloses a brain function state analysis method and device, which comprises the following steps: collecting a brain image sequence combination of a target object, preprocessing, then performing feature extraction on the brain image to obtain image features, and selecting key image features; performing quantitative analysis on the key image features to generate a feature parameter set; inputting the feature parameter set into a brain function state prediction model to determine the brain function state of the target object; the brain function state prediction model is obtained by integrating and then training a semantic retrieval model and a fine-tuning generation model, the semantic retrieval model is trained based on a brain knowledge document library, and the fine-tuning generation model is trained based on reinforced learning parameters, brain information segments retrieved by the semantic retrieval model, and a labeled data set of a mapping relationship; statistical characteristic values corresponding to each brain function state are determined, and brain function physiological data of the target object is analyzed. The application can quantitatively and standardize overall analysis of the brain function state.

The present disclosure relates to a kind of double-layer metasurface image processing method, device, equipment and system.The method comprises: when decryption, using the bottom layer metasurface unit in double-layer metasurface, with the electromagnetic wave of first frequency, to the multiple first password book images obtained from different terminals, respectively, carry out restoration processing, obtain multiple second password book images;Using the bottom layer metasurface unit and cover layer metasurface unit in the double-layer metasurface, with the electromagnetic wave of second frequency, to the first key image obtained, carry out restoration processing, obtain second key image;According to the multiple second password book images and the second key image determine second original image.The above technical solution of the present application adopts two different frequencies to determine second password book image and second key image respectively.It is favorable to increase cracking difficulty and improve security.

The present invention discloses an image processing method for automatic welding of transmission tower components, which belongs to the technical field of tower production. The processing method comprises: obtaining an initial image captured during the automatic welding process, performing channel expansion on the initial image to obtain a first-order feature matrix; using multiple feature generation modules to process the first-order feature matrix to obtain a second-order feature matrix; inputting the second-order feature matrix into an image reconstruction module, and outputting an improved image. The present invention creatively designs a semi-dense inner cross feature extraction mechanism in the first subunit, which can better identify and remove other noise interference while retaining key image information such as welds. The second subunit effectively models the importance of each image area through its internal view transformation calibration mechanism, integrates the first result with the second result, and enables the feature generation module to have a more accurate and robust learning ability for various information.

The invention relates to the technical field of image processing, and discloses a target detection method and device, equipment and a medium. The method comprises the following steps: acquiring a to-be-detected panoramic image; based on the key image features in the to-be-detected panoramic image, performing adaptive slicing on the to-be-detected panoramic image to obtain a plurality of original slice images; performing neighborhood information supplement on the original slice image to obtain a plurality of supplement slice images; performing target detection on the supplementary slice image to generate a corresponding target detection frame; and performing global consistency verification on the target detection frames to remove the target detection frames which do not meet a preset consistency condition, and obtaining a target detection result corresponding to the to-be-detected panoramic image. The embodiment of the invention can be suitable for various panoramic image slice scenes, and the panoramic image slice quality is improved.

The present disclosure provides a key point detection method and device, a computer-readable medium, and an electronic device, relating to the field of artificial intelligence technology. The method includes: obtaining a current image frame and determining key image information corresponding to the current image frame; determining a previous image frame corresponding to the current image frame and obtaining historical key point information corresponding to the previous image frame; inputting the current image frame, key image information, and historical key point information into a pre-trained continuous frame key point detection model, and outputting the current key point information corresponding to the current image frame. The present disclosure can better utilize the spatial and temporal information of video image frames to perform more stable facial key point detection, effectively improving the accuracy and robustness of facial key point detection results in videos.

The embodiment of the invention discloses a certificate anti-counterfeiting detection method, which comprises the following steps of: identifying a certificate type of a certificate corresponding to a certificate image to be detected, marking a key image area in the certificate image to be detected by adopting a mark mask matrix preset for the certificate type as a mark image; and inputting the marked image and detection prompt information preset for the certificate type into the VLM, so that the VML detects whether the certificate image to be detected is a forged certificate image or not according to the marked image under the guidance of the detection prompt information. According to the method, the detection prompt information comprises the first text information used for guiding the VLM to improve the attention weight of the marked key image area in the marked image, so that the attention of the VLM can be guided to the high-risk key image area, and the VLM is prevented from wasting computing power in other unimportant image areas; therefore, the counterfeit certificate image can be detected more accurately, and the security of online business based on eKYC is effectively improved.

The invention discloses a second-hand vehicle copywriting generation method, and the method comprises the steps: obtaining vehicle structured information, feature label information and a vehicle source image set after obtaining a vehicle source identifier; determining key images of description parts such as appearances, interiors and power from the image set; aiming at each description part, constructing part prompt information which contains task constraint, content constraint and length constraint and is injected with structured information, calling a visual understanding generation service based on the key image and the part prompt information to generate a part copywriting, and cleaning the part copywriting; and fusing the structured information, the characteristic label information and the part copywriting to generate a summary copywriting, and outputting a structured copywriting object comprising a summary field and a part field. Through the mode, the source of the copywriting content is limited, part-by-part constrained generation and structured output are realized, the image-text consistency and controllability are improved, assume and quality fluctuation are reduced, and the method is suitable for batch generation.

The embodiment of the invention provides an image recognition method and device, equipment, a medium and a product, and relates to the field of image processing, and the method comprises the steps: firstly segmenting an image into blocks, coding the blocks into embedded vectors, and calculating the contribution degree of each image block to recognition; and then only selecting the key image blocks with the highest contribution degree in the first preset number for weighting and inputting the key image blocks into the recognition model. According to the technical scheme, the model does not need to carry out indiscriminate dense calculation on all regions of the whole image, but can dynamically focus main calculation resources on the key region with rich information according to the image content, so that the calculation load and memory occupation during high-resolution image processing are reduced; by introducing a dynamic selection mechanism based on the importance scores of the image blocks, the problem of low image recognition efficiency in the prior art is solved, and the recognition efficiency is improved while the recognition precision is ensured.

The application discloses a natural water fish monitoring method and system based on image recognition, and relates to the technical field of image recognition, and comprises the following steps: determining a target monitoring fish, and configuring an attracting strategy; executing the attracting strategy, synchronously activating a sonar to execute regional scanning of a target area to establish a time sequence scanning data set, and starting an image acquisition device to perform positioning image acquisition, to establish a time sequence acquisition data set; simultaneously, updating sonar acquisition parameters to synchronously execute sonar update data acquisition, to establish an update data set; after extracting key image frames from the time sequence acquisition data set, performing image processing and recognition on the key frame images, to establish a first recognition result; performing positioning recognition on a time sequence window, to establish a second recognition result; after performing alignment and fusion of the first recognition result and the second recognition result, outputting a fusion recognition result. The application solves the technical problems that fish monitoring in the prior art is susceptible to environmental interference and has low recognition accuracy, and achieves the technical effect of improving fish monitoring accuracy and stability.

The present disclosure discloses a state estimation method, device, and electronic device for a visual inertial odometry, the method comprising: obtaining a current image frame and inertial measurement data; when the current image frame is a key frame, adding the current image frame to a set of key image frames determined based on a sliding window; determining a first inertial residual based on the inertial measurement data; updating a first inertial constraint and a first visual constraint based on the first inertial residual and the first visual residual; estimating the visual inertial odometry state corresponding to the key image frame within the sliding window through iterative optimization; in each iteration, determining incremental change data of the visual inertial odometry state of the current iteration process relative to the previous iteration process based on marginalized prior constraints, the first visual constraints, and the first inertial constraints, and optimizing the visual inertial odometry state based on the incremental change data.