285 results about "Key images" patented technology

The invention discloses a method and device for generating medical image question and answer thinking chain data in multiple stages. According to the method, by introducing a multi-stage model collaborative reasoning mechanism, extraction of key image content, enhanced generation of image description and construction of a thinking chain are completed in sequence, so that the accuracy, pertinence and interpretability in a medical image question and answer task are improved; the problems that an existing model is inaccurate in image detail recognition, opaque in reasoning process, generalized in question and answer content and the like are solved.

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 a music recommendation method, system and device based on user emotion monitoring, and relates to the technical field of music recommendation, and the method comprises the steps: monitoring a music playing instruction in response to the emotion of a user, and collecting a face image sequence of the user; key face image extraction is carried out, a key face image sequence and a non-key image sequence are obtained, and user basic emotion recognition is carried out according to the key face image sequence; performing ambient light analysis according to the non-key image sequence, and performing user emotion correction to obtain a corrected emotion level; and performing music recommendation according to the corrected emotion level. According to the method and the device, the technical problem that the music recommendation result cannot really match the user demand due to the fact that the emotional state of the user cannot be quickly and accurately recognized in the traditional method can be solved; the efficiency, accuracy and reliability of user emotion recognition can be remarkably improved, then music meeting the requirements of a user can be accurately recommended according to the emotion state of the user, and the user experience is improved.

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.

A computer-implemented method for generating a three-dimensional (3D) digital model from one or more two-dimensional images is described. The method includes: obtaining, though an application programming interface (API), a series of two-dimensional (2D) images of a scene taken by an image capturing device; extracting, by a processing device, key images from the series of 2D images, wherein each of the key images depicts one or more components of a building structure in the scene; determining, by the processing device, and based on the extracted key images, a respective position and a respective direction of the image capturing device relative to each of the one or more components of the building structure; and processing, using a 3D image generation neural network, the extracted key images and the positions and directions of the image capturing device to generate metadata comprising a three-dimensional (3D) digital model of the building structure.

The invention belongs to the technical field of industrial safety monitoring, and provides a thermal power plant field operation safety identification alarm system, which comprises image acquisition equipment, network transmission equipment, an image identification unit and monitoring terminal equipment, the image acquisition equipment is used for acquiring video data of field operation; the image recognition unit is used for converting the video data into to-be-input image data and extracting key image data in the to-be-input image data; inputting the key image data into a target detection model to obtain an image recognition result; if the image recognition result is that the personnel violation behavior and / or the environment hidden danger state are included, determining an abnormity type, occurrence time, an occurrence place and image evidence; generating alarm information, and sending the alarm information to a monitoring terminal device; and the monitoring terminal equipment is used for displaying the alarm information and forwarding the alarm information to the mobile terminal equipment of the remote management personnel. According to the system provided by the invention, the identification precision, efficiency and convenience of the field operation safety of the thermal power plant are improved.

The invention relates to the technical field of image processing, in particular to a cultivated land monitoring method and system based on an unmanned aerial vehicle. The method comprises the following steps: obtaining adjacent key image frames of unmanned aerial vehicle video data, and splicing the adjacent key image frames to obtain a panoramic image; performing panoramic segmentation on the panoramic image and obtaining masks reflecting different types of cultivated land through an SEEM model; and extracting the contour of the mask, converting the extracted contour into coordinates in a geographic coordinate system, and mapping each coordinate point with the land category to which the coordinate point belongs. According to the application, farmland monitoring can be carried out by using the aerial image of the unmanned aerial vehicle, the farmland is monitored in real time, meanwhile, image processing is carried out by using the panoramic image splicing technology and the SEEM model, a wider view can be obtained, the limitation of a single image is eliminated, and refined management of the farmland is realized.

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.

The invention relates to a point cloud and finite element model generating system and method based on computer vision, the system comprises a video collecting and processing module, a three-dimensional point cloud establishing module and a finite element model generating module, and the method comprises the following steps: collecting and processing video data of a target structure to obtain a key picture and corresponding spatio-temporal information; performing three-dimensional point cloud reconstruction on the target object by adopting a multi-view angle geometric reconstruction technology based on deep learning based on view angle differences among the plurality of key pictures and corresponding spatio-temporal information, and preprocessing point cloud data; the method comprises the following steps: preprocessing point cloud data, extracting effective structure space geometric information from the preprocessed point cloud data, cutting the point cloud into a plurality of thin layers by adopting a mode of tomography nodes, establishing in a layered manner to obtain finite element components, and integrating the finite element components to obtain a finite element model. Compared with the prior art, the method has the advantages that the video data can be efficiently and accurately converted into the finite element model required by structural analysis, and tedious operation of manual modeling is avoided.

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 utility model discloses a display module and a household appliance display panel with double-layer diaphragms matched for display, and belongs to the technical field of display panels, the display module comprises a first diaphragm and a second diaphragm, the first diaphragm is located on the lower layer in the light source projection direction, the second diaphragm is located on the upper layer in the light source projection direction, and after the first diaphragm and the second diaphragm are matched with a light source, the first diaphragm and the second diaphragm are matched. The display module only displays the background image-text and the function key image-text. According to the utility model, the two diaphragms, the two groups of light guide plates and the double-layer diaphragm are matched to realize the display of images and texts, the upper light guide plate and the diaphragm realize the slight brightness and constant brightness of background light, and the lower light guide plate realizes the highlight of the images and texts corresponding to the area when the touch key area is operated, so that the operation of a user is more convenient. The background light sources and the touch key light sources on the upper layer and the lower layer are independently controlled, and the light guide plate is arranged in a double-layer mode, so that background images and texts and function key images and texts do not affect each other, and display of the background images and texts and the function key images and texts is attractive and striking.

The invention discloses a multi-sensor fusion equipment state estimation method and device, and relates to the technical field of state estimation. In multi-sensor fusion equipment state estimation, point cloud data and image data are subjected to deep fusion in a data level, if a certain point cloud voxel lacks historical observation in a global visual map, prior space information provided in a point cloud map is utilized to obtain an adjacent point set of the point cloud voxel, and the adjacent point set is subjected to image fusion in the global visual map. According to the method, a candidate point set which may be incrementally updated to a global visual map is formed, then the candidate point set is projected to current image data, and key image blocks and corresponding point cloud voxels in the current image data are determined according to angular point response values of projection results, so that the point cloud voxels of the key image blocks in the current image data are determined for the situation of lack of historical observation. And forming an incidence relation between the point cloud voxels with the effective key information and the image blocks so as to update the global visual map. The correlation success rate of the image and the point cloud is remarkably improved, and the state estimation precision is improved.

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.

The invention relates to an abnormal behavior recognition method and system in combination with laboratory video data analysis, and relates to the field of video recognition, comprising: performing key image extraction on a plurality of part monitoring image sequences to obtain a plurality of key part monitoring image sequences, performing key timestamp extraction on a monitoring timestamp sequence to obtain a plurality of key part monitoring image sequences; obtaining a plurality of key monitoring timestamp sequences; performing action recognition according to the plurality of key part monitoring image sequences to obtain a plurality of behavior recognition results, and calculating according to the plurality of key monitoring timestamp sequences to obtain a plurality of change speeds; according to the multiple behavior recognition results and the multiple change speeds, the abnormal behavior probability is obtained through calculation, and a recognition result is obtained through judgment. Through the method, the problems that potential dangerous behaviors cannot be quickly and accurately identified in a dynamic and complex laboratory environment, response is slow and identification accuracy is low in a traditional method can be solved; the efficiency and accuracy of abnormal behavior recognition can be remarkably improved, and potential safety hazards are effectively prevented.

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 belongs to the technical field of image processing, and particularly relates to an image tampering protection method based on a steganography technology, which comprises the following steps of: S1, acquiring an image, inputting associated information, marking a key area on the image to obtain an initial image, and enabling the key area to be an area containing key information; s2, segmenting the initial image into a key region image and a non-key image, inputting the key region image and the non-key image into a steganography module, respectively embedding associated information into the key region image and the non-key image by the steganography module through a steganography technology, and combining the key region image and the non-key image into a transmission image; and S3, decomposing the transmission image through a steganography analysis module to obtain steganography information respectively positioned in the key area image and the non-key image, comparing differences between the steganography information at different positions, and judging whether the transmission image is tampered or not according to a difference value and an identification result. According to the method, the associated information is secretly transmitted through the image, and whether the image is tampered or not can be better judged.

Example systems, devices, media, and methods are described for presenting an interactive game in augmented reality on the display of a smart eyewear device. A hand tracking utility detects and tracks the location of hand gestures in real time, based on high-definition video data. The detected hand gestures are compared to a library of hand gestures and landmarks. Examples include synchronized, multi-player games in which each device detects and shares hand gestures with other devices for evaluation and scoring. A single-player example includes gesture-shaped icons presented on a virtual scroll that appears to move toward an apparent collision with corresponding key images, awarding points if the player's hand is located near the apparent collision and the detected hand shape matches the moving icon.

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.

According to the VR game equipment display control method and system based on the real-time image, the shooting state of the shooting unit on the external environment is adjusted based on the wearing posture of the VR game equipment, the shot external environment image is rendered into the virtual environment image, and the virtual environment image and the external environment have the high compatibility matching degree; based on the real-time game scene attribute information, processing the virtual environment image into a virtual game scene image, determining a key image area and a non-key image area contained in the virtual game scene image, and performing image content identification on the virtual game scene image; and the projection state of the scene image is adjusted in a targeted manner based on the optimal game image watching area of the wearer, so that the key image area and the optimal game image watching area meet the preset relative position condition, and the image display verisimilitude of the VR game equipment and the watching immersion of the wearer are improved.

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 present invention discloses a method and device for automatically generating code for an image interface based on dual attention. DSL is used to describe the GUI, image features are extracted based on Resnet50, and context vectors are extracted using a two-layer LSTM. The image features and context vectors are jointly input into a dual attention module to calculate channel attention and spatial attention respectively, enabling the network to perform weighted processing on the features of the input image from multiple aspects. The generation layer combines the attention output to classify the score tensor, and the DSL sequence output by the model is finally compiled by the Compiler into source code for a specific platform. The dual attention mechanism that fuses the spatial attention mechanism and the channel attention mechanism proposed by the present invention can provide the spatial position of the corresponding GUI component at the current time step when the decoder learns the DSL variable relationship, and allocate effective attention to the key image feature maps according to the calculated channel importance degree, greatly enhancing the network learning ability and improving the accuracy of generating code for the image interface.

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.