89 results about "Image mode" patented technology

The invention belongs to the technical field of geological disaster risk assessment, and particularly relates to a slope stability grade identification method and system based on multi-modal deep learning. Comprising the following steps: acquiring a remote sensing image and corresponding parameter data, and respectively preprocessing the remote sensing image and the corresponding parameter data to obtain a high-dimensional image feature vector and a parameter feature vector; performing dynamic fusion of multi-modal information on the high-dimensional image feature vector and the parameter feature vector through a modal-level gating and fine-grained weighting mechanism to obtain a fusion feature; and inputting the fusion features into a preset classifier to obtain probability distribution of slope stability levels, and outputting the class with the maximum probability as a stability level identification result. According to the method, through combined modeling of the image modality and the parameter modality and introduction of an improved gating attention mechanism in a feature fusion stage, a reliable explanatory basis can be provided while high-precision prediction is ensured, so that the engineering applicability and the popularization value of the model are enhanced.

The invention provides a knowledge graph construction method and device, equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring a script text and a deductive video corresponding to the script text; constructing a plurality of first nodes of a text mode based on the script text, and constructing a plurality of second nodes of an image mode and a plurality of third nodes of an audio mode based on the deductive video; based on the content information corresponding to the fourth node and the fifth node, determining a semantic consistency score between the fourth node and the fifth node; determining a time sequence synchronization score between the fourth node and the fifth node based on the time sequence information corresponding to the fourth node and the fifth node; in response to the fact that the semantic consistency score is larger than a first threshold value and the time sequence synchronism score is larger than a second threshold value, constructing a first relationship between the fourth node and the fifth node; and constructing the knowledge graph based on the plurality of first nodes, the plurality of second nodes, the plurality of third nodes and the plurality of first relationships.

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.

The invention discloses a welding seam defect X-ray image intelligent diagnosis and credible traceability method based on multi-modal heterogeneous information collaboration, which is characterized in that a defect analysis network fusing multi-domain feature modeling and graph structure expression is constructed on the basis of bimodal data formed by a welding seam X-ray image and an industry detection standard text. In the image mode, dividing the weld seam image into a plurality of local area units through superpixel segmentation, taking the areas as image nodes, respectively extracting spatial domain, frequency domain, wavelet domain and edge domain features, and constructing a weighted graph structure by combining the spatial adjacency relation and the feature similarity relation between the areas; realizing overall modeling and correlation analysis of weld defect structure information by using a graph convolutional network; in a text mode, feature coding is carried out on an industry detection standard text, and the feature coding is used as an important prior constraint for defect judgment. Collaborative modeling of an image detection result and standard semantic information is achieved through a gating fusion mechanism, a mapping relation between a detection conclusion and a standard term is established, and interpretable expression and result credible traceability of the weld defect diagnosis process are achieved. And a welding seam X-ray film automatic digital acquisition and observation device is adopted in a matched manner, so that stable transmission, positioning observation and high-resolution digital imaging of the industrial ray film are realized, and reliable and consistent image data input is provided for the intelligent diagnosis method. The method is suitable for intelligent defect detection under complex welding seam structures and multi-working-condition imaging conditions, and has high engineering application value and popularization prospect.

The invention discloses a large model image segmentation method and system based on multi-scale fusion and a medium, and relates to the technical field of image processing. Obtaining an image segmentation model; the structure construction mode of the image segmentation model is that a traditional SAM is used as a network basis, an Adapter module is inserted between a multi-head attention layer and a multi-layer perceptron in each VIT block of an image encoder of the SAM, and a four-layer CBR-Net module is arranged at the end of the image encoder in parallel. The output end of the image encoder and the back of an up-sampling layer of a mask decoder of the SAM are respectively connected with an Atte-FFB module in series, and residual connection is constructed between the first three-layer output of the CBR-Net module and the Atte-FFB module of the corresponding layer in the mask decoder of the SAM network; residual connection is constructed between the fourth layer output of the CBR-Net module and an Att-FFB module behind the output end of the image encoder; and performing image segmentation on the to-be-segmented image through the image segmentation model. According to the method, the segmentation precision is improved, and the model generalization ability of a cross-image mode is enhanced.

A question and answer data construction method comprises the following steps: 1) reading samples containing image and text data in batches, constructing a unified instruction for each batch of samples, and calling a large language model to extract candidate questions which have discriminability and teaching value and can be answered from images to form multiple groups of candidate questions; (2) carrying out structured deduplication merging on the cross-batch candidate problems obtained in the step (1) to obtain a deduplication problem pool; 3) for each sample, identifying an available image mode of the sample, screening candidate problems matched with the mode from the problem pool, and randomly extracting candidates of which the number is greater than a target value q; 4) constructing a request only based on image answering, calling the multi-modal large language model by taking the sample image as input, and generating a question-answer pair; if reliable answers of the candidate questions cannot be obtained only through the images, replacing the candidate questions with new candidates and retrying until q question and answer pairs capable of being answered are accumulated and obtained; 5, the generated question and answer pairs are subjected to structured verification and safe disking.The method supports parallel processing and breakpoint continuous running so as to meet the large-scale data construction.The method has the advantages of being good in universality and expandability and standardized in result.

Embodiments of the present application provide a game data processing method and device, electronic equipment and readable storage medium, the method comprises: in the game process, the terminal can obtain game resources, wherein the game resources can include scene resources, model resources and obstacle information for scene resources, model display parameters for model resources, then the corresponding game scene of the scene resources can be displayed in the form of two-dimensional image mode, and the corresponding game model of the model resources can be displayed in the form of three-dimensional image mode in the game scene according to the model display parameters, then the game scene can be processed according to the obstacle information, the corresponding obstacle map is generated, and the moving information of the game model is obtained, and then the target moving path of the game model in the obstacle map is determined according to the moving information, so as to control the game model to move in the game scene.

A Virtual Reality (VR)-based seat control apparatus for a vehicle includes: an input unit for receiving an image mode operation command from a user; and a control unit for playing a prepared timetable according to the image mode operation command, performing interlocking with a seat module of a vehicle and VR equipment, and controlling an operation module of the seat module according to the timetable.

The present disclosure relates to a panoramic image display system, method, vehicle, electronic device and storage medium, and relates to the field of vehicle design, and the system comprises: in the case of starting a panoramic image function and a UAV image auxiliary function, controlling a UAV to take off from a vehicle and fly to a first target position, and acquiring a UAV image from the UAV, the UAV image being acquired by the UAV from the vehicle at the first target position according to an image mode of the UAV image auxiliary function at present, and according to a display mode of the panoramic image function at present, controlling the display screen to display a panoramic image containing the UAV image, through the UAV image auxiliary function, the target position required to acquire the image is acquired by the UAV, and the UAV image is displayed on the display screen, which can reduce image distortion and image blind area, and can provide effective image assistance for users.

The invention relates to an air component analysis method and system in an airport environment, and the method comprises the steps: continuously collecting the wind field motion in the airport environment through a laser wind measurement radar in a preset time period, and obtaining an initial wind field motion point cloud data set; performing wind field displacement vector calculation on each wind field motion point cloud data in the set to obtain a wind-point-free cloud data set; constructing a corresponding triple for each piece of windless point cloud data in the set; and performing normalization and visualization processing on the triple to generate image data, and analyzing the image data through an air component classification model to obtain air components of the airport environment. By means of the air component analysis method and device, the wind field state which changes severely in the airport environment is converted into the windless state, the emission characteristics of pollution sources are reflected more truly, large-range air component analysis is conducted in a triple-image mode, and the problem of how to improve the applicability of air component analysis in the large-range environment is solved.

In order to confirm the activity state of the pelvic floor muscle in practice of the pelvic floor muscle exercise, a sensor sheet is attached to the abdominal region corresponding to the transversus abdominis muscle, and an electromyogram of the transversus abdominis muscle which contracts in cooperation with the pelvic floor muscle is acquired. At this time, even if the transversus abdominis muscle contracts, the pelvic floor muscle does not necessarily contract, and thus the activity state of the pelvic floor muscle is confirmed by echo. In more detail, an ultrasonic sensor is pressed against the abdomen in a manner that an ultrasonic probe is placed on the sensor sheet, and thus an image (M-mode image) of an ultrasonic echo representing the activity state of the bladder bottom which can be considered to be the same as the contraction of the pelvic floor muscle is acquired. Thereby, the implementer of the pelvic floor muscle exercise can experience the knack of contracting the pelvic floor muscle, that is, what kind of consciousness to focus on which part of the body can contract the pelvic floor muscle.

The application belongs to the field of optical storage, and specifically discloses an optical disc data reading method and device. The optical disc involved in the application comprises a plurality of data layers, each data layer comprising a plurality of data blocks. The optical disc data reading method comprises the following steps: focusing imaging light to a preset data layer of the optical disc; obtaining images of each data block in the preset data layer based on the imaging light when the optical disc rotates; controlling the imaging light to stably focus on the preset data layer according to the brightness and the definition of each data block image, and then reading corresponding data information based on the images of each data block. According to the application, the defocus distance and direction can be accurately predicted in combination with the definition evaluation function, and real-time focus control can be performed in the process of continuous rotation of the optical disc. Data is read in a pure image mode, and focus servo control can be completed without making a data recording layer with a track in the optical disc, so that reading of hundreds of layers of data in a multi-dimensional storage optical disc can be realized.

This technology provides a way to switch to video recording mode in response to a half-press operation of the control element while shooting in still image mode. [Solution] The present invention provides an imaging device comprising: a first operating member having a non-pressed state, a half-pressed state, and a fully-pressed state; a storage unit that stores setting values ​​for still image shooting and setting values ​​for video shooting, respectively; and a control means that controls the display means to switch to displaying an image shot using the video setting value when the first operating member transitions from the non-pressed state to the half-pressed state while an image shot using the still image setting value is being displayed on the display means, and to start recording a video shot using the video setting value when the first operating member transitions to the fully-pressed state.

The invention relates to the technical field of diaphragm imaging, in particular to an imaging mode management system and method based on a porous diaphragm, and the system comprises an angle deviation extraction module which determines hole site transition and direction logic, a homodromous kinetic potential analysis module which extracts a consistent rotation action sequence, and a hole site switching instruction construction module which generates a coherent driving instruction. The offset trend extraction module collects physical offset readings and constructs a holding section graph, and the mode management generation module matches a compensation command and drive scheduling and generates a porous diaphragm imaging mode management set. According to the method, a clear basis is provided for imaging compensation by combing the corresponding relation between the hole site and the rotation direction, unifying the direction sequence and the execution sequence, determining the clear transmission logic and the continuous connection characteristic in the switching process, keeping the same-direction continuity of multi-hole switching, reducing the control uncertainty, integrating the path and the physical offset data and identifying the offset change keeping section; rotation predictability is maintained under complex conditions, and imaging mode scheduling stability and control precision are improved.

The application discloses an image classification method and device, a storage medium and a computer device. The method is applied to the computer device and comprises the following steps: in a self-defined image classification mode of an electronic album, in response to a triggering operation of a user on an update control of an image corresponding to a to-be-managed image category, updating the image in the to-be-managed image category, calling a training interface, obtaining a first image classification model corresponding to the self-defined image mode through the training interface, the model being pre-trained, and performing transfer training on all images of the to-be-managed image category according to the first image classification model, so as to obtain a first image classification model after transfer training. The classification of the updated image can be matched by using the first image classification model after transfer training, the classification accuracy is improved, finally, the to-be-classified images of the electronic album are classified by using the trained first image classification model, the image classification efficiency and accuracy are improved, and the image searching efficiency is further improved.

The utility model provides inspection equipment. The inspection equipment comprises a scanning channel; the first scanning device is configured to radiate rays towards the to-be-detected object of the scanning channel so as to form a first scanning image; the second scanning device is arranged on one side of the first scanning device and configured to radiate rays towards the to-be-detected object to form a second scanning image, and the first scanning device and the second scanning device have different imaging modes; a first accommodating space is formed in the first cover body, and a first scanning device is accommodated in the first accommodating space; a second accommodating space is formed in the second cover body, and a second scanning device is accommodated in the second accommodating space; and the heat dissipation device is arranged on one of the first cover body and the second cover body. According to the inspection equipment, heat dissipation of the first scanning device and the second scanning device according to needs is achieved through the heat dissipation device, the heat dissipation requirements of the two scanning devices can be met, and meanwhile energy consumption is low.

The invention discloses a multi-modal depression auxiliary detection algorithm, and the method comprises the steps: obtaining multi-modal data which comprises video data, voice data and scale data of a medical site; carrying out identity recognition, round labeling and format conversion processing on the multi-modal data to obtain structured data containing role information, time sequence round and content fields; performing feature extraction on the structured data to obtain a feature vector set containing an image mode, a text mode and a voice mode; carrying out fusion processing on the feature vector set by adopting a preset cross-modal attention mechanism dominated by a text mode to obtain multi-modal fusion feature information; and inputting the multi-modal fusion feature information into a pre-configured depression risk assessment model, performing feature transformation and compression through a full connection layer and a pooling layer, and outputting a depression disease probability for representing a target individual. According to the method, complementation and dynamic semantic modeling of multi-source data can be realized, and the accuracy and reliability of a detection result are improved.

The embodiment of the invention relates to an image cross-modal saliency detection method and device, equipment and a medium. The method comprises the steps that a color image and a depth image corresponding to the color image are acquired; determining a first phase diagram corresponding to the color image, and determining a second phase diagram corresponding to the depth image; performing saliency phase updating processing based on preset distribution according to the color image, the depth image, the first phase diagram and the second phase diagram to obtain a saliency phase diagram; and according to the color image and the depth image, carrying out conversion from a frequency domain to a space domain on the saliency phase diagram to obtain a saliency map. According to the embodiment of the invention, on the basis that the discovered color image and the depth image have phase consistency of frequency spectrums with the saliency map, that is, the phase-related data obeys the preset distribution, the two modes of color and depth are taken as input, the saliency map is taken as output, and three different image modes are involved; and cross-modal saliency target detection with interpretability and relatively high accuracy is realized.

The invention discloses a data processing method and device, a storage medium, equipment and a program product, which are applied to a data processing scene. The method comprises the following steps: obtaining query data, wherein the query data comprises at least one modal data of a text modal, an image modal and an audio modal; based on the query data, a query feature vector of the target feature dimension is generated through a multi-modal retrieval model, and the multi-modal retrieval model is configured to be capable of outputting feature vectors of multiple different feature dimensions; and performing similarity calculation on the query feature vector of the target feature dimension and a candidate feature vector corresponding to each candidate data in the candidate data set to obtain a retrieval result. According to the method, by maintaining the multi-modal retrieval model, the output of the feature vectors of different feature dimensions is supported, the resource consumption is reduced, the flexibility is improved, and the retrieval performance can be ensured.

Eyewear having an image signal processor (ISP) dynamically operable in a camera pipeline for augmented reality (AR) and computer vision (CV) systems. Multi-purpose cameras are used for simultaneous image capture and CV on wearable AR devices. The cameras are coupled to a frame and configured to generate images, wherein the cameras and the ISP operate in a first AR mode and capture images having a first resolution suitable for use in AR, and are configured to operate in a second CV mode to provide the images having a second resolution suitable for use in CV. The first resolution in the AR mode is higher than the second resolution in the CV mode, and the cameras and the ISP consume less power in the second CV mode than the first AR mode. The cameras and the ISP save significant system power by operating in the low power mode CV mode.

To provide a method for easily visualizing a latent image without requiring a discrimination tool such as a parallel line filter and a lenticular lens.SOLUTION: By a terminal including at least a display unit, an image processing unit, and any one of a storage unit and an imaging unit, from image data of a latent image pattern including a plurality of latent image elements in which a latent image is visualized by being periodically sampled, A latent image visualization method for visualizing a latent image hidden in a latent image pattern includes a step of displaying image data of the latent image pattern captured by an imaging unit or image data of the latent image pattern stored in advance in a storage unit on a display unit, and a step of reducing the image data of the latent image pattern by an image processing unit, thereby visualizing the latent image on the display unit.SELECTED DRAWING: Figure 6

Provided are a refreshing method and a display apparatus. The method includes: presenting a video on a display of the display apparatus; detecting whether a content type of the video is switched from a first content type to a second content type; obtaining first picture quality parameter information matching the second content type from preset data of the display apparatus, wherein the preset data includes a predefined relation between a content type and picture quality parameter information, updating the parameter of the at least one picture quality setting option in the picture mode settings menu according to the first picture quality parameter information, and presenting the second section of the video according to the updated parameter of the at least one picture quality setting option in the picture mode settings menu.

Arrangements for detecting polymorphic botnets using a pattern matching engine are provided. The platform may train a pattern matching engine to generate image pattern statuses based on a training set of historical image patterns. The platform may automatically generate real-time interaction information based on monitoring interactions between user devices and cloud platforms. The platform may generate a potentially malicious image pattern based on the real-time interaction information. The platform may then input the image pattern into the pattern matching engine to generate an image pattern status for the image pattern. The platform may store the image pattern to a pattern storage repository. Based on identifying the image pattern as malicious, the platform may initiate a security action. The platform may update the pattern matching engine based on the image pattern and its image pattern status to detect future polymorphic botnet attacks.

The application discloses a kind of fast high-resolution imaging methods based on vortex electromagnetic wave, which constructs the two-dimensional imaging model of pitch angle and azimuth angle based on concentric circular ring array, realizes low-rank decomposition to the received echo and observation matrix by means of spherical harmonics function, introduces orthogonal matching pursuit algorithm to the scattering point in observation plane for target reconstruction, so as to realize the two-dimensional joint imaging of radar target pitch angle and azimuth angle.The imaging mode of joint OMP algorithm and spherical harmonic decomposition proposed in the application can realize higher imaging resolution compared with the traditional generalized inverse solution mode, while greatly reducing the operation amount of imaging process, effectively saving calculation time.

An X-ray CT apparatus according to an embodiment includes a scanner section, a support section, a selection section, and a control section. The scanner section has an imaging system. The support section has a movement mechanism for the scanner section, and supports the scanner section. The selection section selects a unit for moving the scanner section according to an imaging mode based on the state of a subject. The control section controls the unit according to a driving condition for each imaging mode.

The present application relates to the technical field of small sample image classification, in particular to a small sample classification method based on multi-granularity image-text prototype matching of task-driven structure; the method comprises the following steps: constructing a support set and a query set; inputting the two sets into a multi-granularity feature set extraction network; inputting an image mode into a multi-level image feature extraction network composed of an improved ResNet12 and an embedded task-driven mapper selection module to extract low, medium and high level image features; calling a large language model LLM three times in sequence to generate word-level, sentence-level and paragraph-level semantic descriptions, and combining a text editor to form multi-angle text features; inputting the image and text feature sets into a semantic graph interaction fusion network to realize multi-level information fusion through node and edge construction, cross-modal attention, graph convolution and gating mechanism; calculating matching scores of each layer through a multi-granularity prototype matching network to perform classification prediction; and adopting a multi-granularity compound weighted loss function to optimize the model performance in the training stage.

A kind of detection method and detection equipment of flux distribution condition can be used to show the flux of electronic component surface.For the bump area of electronic component, according to the predetermined thickness of the applied flux, to select the first bump area imaging mode with low angle of light irradiation or the second bump area imaging mode with high angle of light irradiation.The first bump area imaging mode adopts the imaging angle of 80-100 degrees with the carrier plate of the electronic component to obtain the image.The second bump area imaging mode adopts the imaging angle of 60-70 degrees with the carrier plate of the electronic component to obtain the image.For the non-bump area of electronic component, high angle of light irradiation is used and the imaging angle of 80-100 degrees is used to obtain the image.In this way, the flux distribution of bump area and non-bump area can be correctly shown under different detection conditions.

The application discloses a kind of image-text classification method, system and equipment, it is related to computer vision, natural language processing and other technical fields.Image-text classification method includes: original image data and text description data are respectively encoded and handled, and image feature data and text feature data are obtained;Image feature data and text feature data are spliced and handled, and spliced feature data is obtained;Spliced feature data is randomly discarded and handled, and image-text feature data is obtained;Image mode and text mode in image-text feature data are multi-modal complementary fusion processing, and target fusion feature data is obtained;Target fusion feature data is classified and handled based on multilayer perception, and image-text classification result is obtained.

Multimodal radar antenna device (100, 200, 300), comprising: a transmitting antenna section (120, 220, 320) comprising a first plurality of transmitting antennas (101) configured to transmit a steerable mode signal, a second plurality of transmitting antennas (104) configured to transmit an imaging mode signal, and a third plurality of transmitting antennas (103) configured to transmit the imaging mode signal and the steerable mode signal; and a receiving antenna section (140, 240, 340) comprising a plurality of receiving antennas (102), wherein the second plurality of transmitting antennas (104) is arranged between the first plurality of transmitting antennas (101), wherein the second plurality of transmitting antennas (104) is arranged between the third plurality of transmitting antennas (103), wherein the first plurality of transmitting antennas (101) comprises a first set of steerable mode signal transmission antennas (121) with a spacing of 1.5 λ between antennas of the first set of steerable mode signal transmission antennas (121) and a second set of steerable mode signal transmission antennas (122) with a spacing of 1.5 λ between antennas of the second set of steerable mode signal transmission antennas (122), wherein the first sentence and the second sentence are arranged horizontally along an axis, and where λ is a wavelength of a signal that is sent or received by an antenna.

The application discloses a method for improving color accuracy of a liquid crystal display, which is characterized in that a backlight device of the display is matched, a body color temperature and a luminance characteristic value are designed, a preheating time of the display is set, after the display is started and waits for the preheating time, maximum luminance cycle detection of the whole machine, body color coordinate cycle detection of the whole machine, automatic algorithm debugging of a Gamma multi-stage wide coefficient range, accurate color temperature debugging, sRGB and DCI-P3 color space matrix conversion are carried out in a darkroom environment, then a plurality of color pictures are outputted for color accuracy test, an average color accuracy value is calculated, parameters corresponding to color accuracy values meeting preset requirements are written into the display, and the parameters are called by different color space image modes. The application quantitatively designs liquid crystal backlight luminance, color temperature and other characteristics, then accurately calculates and debugs the display online, corrects color temperature, color gamut overlap, white balance, Gamma and other parameters affecting the color accuracy value, carries out real-time color space conversion according to the measured characteristics of the display screen, adjusts color data and luminance data of each display in real time, so that the color restoration of the liquid crystal display is improved, the color accuracy value of each display is reduced, and the color accuracy is improved.