389 results about "Image texture" patented technology

The invention discloses a few-sample defect identification method based on cross-modal text semantic driving, and belongs to the technical field of image processing. Aiming at the problem of insufficient generalization of a detection model caused by scarcity of abnormal samples and dynamic evolution of defect types in an industrial quality inspection scene, an unknown defect type can be accurately identified only by a small amount of normal data by establishing a dynamic feature recombination mechanism and an adaptive discrimination boundary; according to the method, a simulation sample similar to a real defect in form is generated on a normal sample through a matching relation between text description and image features; when a defect type which is not seen is encountered, a comparison standard of image textures can be automatically adjusted according to text semantics, subtle differences between a normal area and an abnormal area can be accurately distinguished, dependence on real defect data is not needed, and the sample defect identification precision is further improved.

The invention discloses a water quality monitoring method and system based on artificial intelligence. The method comprises the following steps: acquiring a comprehensive data set composed of sensor data, satellite images and meteorological parameters; according to the water flow velocity and pollution concentration gradient in the comprehensive data set, adopting a dynamic sampling algorithm to adjust the sampling frequency and position, and outputting adjustment data; performing space-time interpolation processing on the adjusted data to obtain a preprocessed data set with high-density space-time coverage; key features of sensor values, image textures and meteorological parameters are extracted from the preprocessed data set by adopting a principal component analysis method, a weighted feature matrix is constructed, and a fusion feature set is obtained; and judging whether the dimension of the fusion feature set exceeds a preset threshold value, if the dimension of the fusion feature set exceeds the preset threshold value, performing dimension reduction and classification on the fusion features by adopting a random forest algorithm, and optimizing model parameters through cross validation to obtain a pollution concentration prediction result. Effective technical support is provided for water environment protection, and important ecological and social benefits are achieved.

The invention discloses a tunnel disease recognition model training method and system based on a point cloud and an image, and the method comprises the steps: synchronously collecting tunnel point cloud and image data, carrying out the calibration and registration, achieving the spatial alignment, preprocessing the point cloud, generating a gray-scale image and a depth image, collecting continuous images through a line-scan digital camera, generating a spliced image, and carrying out the recognition of tunnel diseases. Partitioning a large-size image after multi-image space-time synchronization; a multi-branch network is constructed, point cloud geometry and image texture features are extracted by using Point Net / 3DCNN and CNN / Transform respectively, and semantic collaborative fusion is realized through an intermediate layer fusion module; parameters are adjusted according to errors through self-adaptive training, manual labeling dependence is reduced in combination with transfer learning and the like, and convergence is accelerated through online iteration, joint loss and self-adaptive weight to improve generalization; the performance of the model is evaluated through field testing and indexes in various tunnel environments, and the structure is optimized according to data to ensure that engineering is feasible and efficient. The model can accurately identify various diseases on different tasks, and can adapt to complex and changeable working conditions in tunnel detection.

The invention discloses a coal bunker reserve real-time monitoring method integrating a laser radar and a camera. The method comprises the following steps: step 1, carrying out space-time joint calibration and hardware-level synchronization on a laser radar and a camera, and establishing a projection relationship between a point cloud and an image; 2, cooperatively collecting coal bunker data based on the calibration parameters, and obtaining a point cloud and a high-resolution image; step 3, heterogeneous preprocessing is carried out on the collected point cloud and the image, denoising, segmentation and down-sampling are carried out on the point cloud, semantic segmentation is carried out on the image, and coal and foreign matters are identified; step 4, based on image texture and point cloud coordinate coupling mapping, reconstructing a three-dimensional model with a semantic tag, projecting the point cloud to an image plane to perform feature matching so as to endow the texture, and fusing image semantic information to correct the model; and 5, dynamically calculating the volume of the coal pile by using the semantic three-dimensional model to realize real-time monitoring of reserves. The system overcomes the limitation of a single sensor, achieves the real-time and high-precision monitoring of the reserves of the coal bunker, and effectively improves the intelligent level of coal bunker management.

The invention discloses a non-aligned spatio-temporal data driven multi-sensor fusion method and system, and the method comprises the steps: building a matching measurement function through collecting radar point cloud density and visual image texture, and constructing a synchronous gradient field through employing a space depth change parameter and a visual motion contour, and determining a data synchronization standard; detecting a time sequence abnormal accumulation area based on a synchronization standard, and identifying an abnormal starting point and propagation time delay through a backtracking track to form an asynchronous sensor coordinate table; iMU angle increment and sensor response delay are collected, and sampling imbalance information is generated through time domain matching of an attitude switching signal and a data collection inflection point; performing association mapping on the asynchronous coordinate table and the sampling imbalance information to generate a feature offset, determining a matching core area, and creating a hierarchical calibration sequence; performing internal and external source difference identification analysis on the calibration sequence to obtain an offset transfer speed to form an offset coupling factor; and a fusion enhancement region is determined based on the offset coupling factor, and high-precision multi-source data fusion is realized through cascade analysis and gain coefficient coding.

The invention discloses a roadside guardrail deformation detection method based on image and point cloud fusion and related equipment, and the method comprises the steps: collecting a guardrail region image through a high-definition camera carried by an unmanned plane, and synchronously collecting the point cloud data of a guardrail region through a laser radar; then, a joint calibration and iterative nearest algorithm is used for carrying out space-time registration on the guardrail area image and the point cloud data which are synchronously collected, and a multi-modal data set is generated, so that the problem of space-time dislocation caused by sensor movement is solved; an improved double-branch deep learning model is adopted to respectively extract image texture features and point cloud geometric features from a multi-modal data set, and the two features are fused to overcome a single-modal defect; and extracting the contour of the guardrail beam plate from the fusion feature map, aligning the contour of the guardrail beam plate with the reference model, and carrying out quantitative calculation to quantify the deformation degree of the guardrail and realize quantitative detection of the deformation of the guardrail.

The invention discloses a marine plankton hyperspectral imaging detection system, relates to the technical field of spectrum detection, and is used for solving the problem of poor spectrum classification and recognition in a water body environment. According to the method, the hyperspectral image and the environmental disturbance parameters are synchronously acquired, and the disturbance mapping sequence is constructed, so that accurate alignment between the image and disturbance is realized. A disturbance contribution weight matrix is constructed based on a dominant wave band, pixel-level spectrum stripping is carried out, a plankton purification spectrum is extracted, the spectrum purity and the recognition accuracy are improved, and a spectrum abnormal drift region is precisely recognized in combination with derivative spectrum change and image texture features; a coupling relation between the drift region and background disturbance is further established, a dynamic interference factor map is generated, and identification model parameters are dynamically adjusted, so that the identification stability and classification accuracy of the system in a complex interference environment are improved, and the method is suitable for multi-scene marine ecological monitoring.

The invention discloses a lingual surface diagnostic information acquisition, diagnosis and analysis method and system, and belongs to the technical field of medical auxiliary diagnos.The method comprises the steps that an acquired lingual surface image is matched with patient information, a symptom associated lingual surface area is determined, the associated lingual surface area is subjected to priority division, and whether the acquired image meets a clear standard or not is judged; the method comprises the steps of constructing an image index evaluation model based on tongue vibration and image texture, performing tongue vibration, texture stability and water vapor fuzzy interference evaluation on an image needing to be processed, constructing a stable frame evaluation model, and importing vibration intensity, texture stability and a water vapor proportion into the stable frame evaluation model to evaluate image area stability. Image registration is carried out on the stable frame set, multi-frame registration and fusion are carried out based on an image stable region, region stability and processing information are recorded, a high-quality image for tongue picture analysis is generated, the image definition of a key diagnosis region is improved, and the accuracy and stability of tongue picture analysis are improved.

The invention relates to the technical field of roadway surrounding rock danger identification, and discloses a roadway surrounding rock danger identification model construction method, which comprises the steps of collecting multi-modal data, and generating preprocessed data through synchronous calibration and denoising; extracting a seismic wave frequency domain and image texture features, and generating a multi-modal feature matrix; in combination with a geological prior clustering mining abnormal mode, generating a labeled sample data set; generating a danger identification model based on a transfer learning and feature fusion training network; and the edge deployment model performs real-time reasoning, and generates an early warning result through an adaptive algorithm. According to the method, the frequency domain features of the seismic fluctuation signals and the depth texture features of the surrounding rock images are fused, the multi-modal feature matrix is constructed, abnormal mode mining is carried out in combination with geological prior knowledge, and early weak abnormal signals such as hidden fault slippage or asymmetric microfracture extension which are difficult to find by a single monitoring means can be effectively recognized.

The invention discloses an image video super-resolution enhancement method based on a degradation generative adversarial network, and relates to the field of image processing, and the method comprises the steps: carrying out the image collection and preprocessing; building and training a super-resolution enhancement model; and carrying out super-resolution enhancement on the image based on the degradation generative adversarial network model. According to the method, an image content self-adaptive dynamic degradation kernel generation mechanism is adopted, the degradation process of the image under different equipment and organization structures is truly simulated, a dynamic up-sampling and residual error correction network guided by the degradation kernel is adopted, the detail reduction capability and the structure fidelity of the super-resolution image are remarkably improved, and the super-resolution image quality is improved. The image texture authenticity and key organization density consistency are effectively enhanced, the balance of training games between a generator and a discriminator is realized, and the model stability and convergence quality are improved.

The invention discloses a water heavy metal concentration simulation and prediction method, which comprises the following steps: integrating original monitoring data, hydrodynamic data, total suspended solids, image remote sensing data and human activity data, and generating a multi-source cleaning sequence data packet; executing cross-modal adsorption capacity estimation by using image remote sensing data in the data packet, inferring particle chemical composition and adsorption isotherm parameters from image textures, and generating a capacity feature packet containing an adsorption capacity upper bound; time-varying travel time is calculated based on the hydrodynamic data and the human activity data, causal alignment is performed on the capacity feature packet and the upstream signal, and a travel time alignment feature packet is generated; and in combination with metal fingerprint parameters, applying an adsorption capacity upper bound as a physical constraint on a form distribution constraint head, explicitly decoupling and predicting the form, and generating a prediction result packet. According to the method, the hydrodynamic physical mechanism and the particle adsorption chemical mechanism are deeply coupled, and the prediction precision and the physical consistency of the model under the unsteady state condition are improved.

The invention relates to the technical field of industrial mechanical arm control, in particular to a mechanical arm execution control method for industrial production, which comprises the following steps: collecting parameters in real time; generating a geometric risk index; generating a mechanical risk index; extracting image features; abnormal mechanical arm judgment; determining an adjusting mechanical arm; and generating an adjustment instruction. According to the method, the inclination angle, the plane offset distance, the contact image features, the tightening torque and the force value of the tail end of the mechanical arm are monitored in real time in a multi-dimensional mode, a dual-risk judgment model based on geometric offset and mechanical loading is constructed, image texture fluctuation and posture change are further combined, the continuously abnormal mechanical arm is accurately positioned, and the safety of the mechanical arm is improved. And geometric and mechanical risk indexes are evaluated again after adjustment, so that the aluminum shell positioning and bolt assembling precision is guaranteed, and the problems that the production efficiency is reduced and the product defect rate is increased due to assembling quality fluctuation caused by mechanical arm positioning errors and inaccurate force control are effectively solved.

The invention provides an underwater image recognition method based on deep learning, and relates to the technical field of information, and the method comprises the steps: preprocessing an original image, recognizing image texture interference caused by the density of suspended particles in an interfered region, separating a noise signal, and obtaining a first processed image; performing deep analysis on the multi-scale feature response values and the key feature points in combination with an underwater target recognition task, and judging the category and position information of the target object to obtain a preliminary recognition result; comparing the preliminary identification result with an actual marine environment condition to obtain a matching degree between the identification result and an expected target feature, and if the matching degree is lower than a preset matching threshold value, adjusting a scale weight parameter and a feature point screening threshold value to obtain corrected identification data; and analyzing changes caused by environmental condition fluctuation according to the corrected identification data, generating target identification output, and determining the accurate position and category information of the underwater target through confidence weighted fusion and coordinate precision correction processing of the identification data.

The invention relates to the technical field of multi-unmanned-aerial-vehicle cooperative operation and three-dimensional modeling, in particular to a multi-unmanned-aerial-vehicle cooperative three-dimensional rapid modeling method for a highway accident scene, and the method comprises the steps: generating a three-dimensional grid map of an accident area through the scanning of a millimeter-wave radar by a main control unmanned aerial vehicle; subareas are divided according to a load balancing strategy and are distributed to slave unmanned aerial vehicles, the slave unmanned aerial vehicles traverse grids along a snake-shaped track, laser radar point clouds and five-view-angle images are synchronously collected, data are bound through double time stamps and space coordinates, the point clouds are preprocessed through edge computing nodes, and the point clouds are stored in a database; the master control unmanned aerial vehicle evaluates quality based on density standard deviation and overlapping matching degree and instructs to reacquire, performs high-precision Poisson reconstruction on an accident core area, performs voxelization processing on a peripheral area, maps image textures, optimizes vehicle deformation details, simplifies a model and retains key element precision, and finally performs data processing. License plate coordinates, a scattered object thermodynamic diagram and an emergency lane occupation state are automatically marked, a visual model is generated, and rapid and accurate restoration of an accident scene is realized.

The invention discloses a watermark imperceptible embedding and recovering method based on a deep learning network structure. Firstly, a mask guide watermark embedding scheme is designed, a mask generation module is constructed by using a residual dense feature extraction module and an attention mask generation module, and a watermark is adaptively guided to be embedded into an image texture rich area so as to improve the invisibility of the watermark. And secondly, constructing a watermark decoding network based on comparative learning, and by comparing a loss function, taking the decoding features of the same watermark image under different noise conditions as positive samples and taking the decoding features of different watermark images as negative samples so as to enhance the consistency of the decoding features, thereby improving the robustness of the watermark. The deep learning network structure can improve the robustness of the watermark in a real screen shooting scene, and has a huge application value in copyright protection and traceability tracking.

The invention provides a cross-domain Internet of Things equipment intelligent collaboration method and system based on a semantic knowledge graph. The method comprises the following steps: adjusting the sampling frequency of a cross-domain Internet of Things device according to a level value, and after the sampling frequency is adjusted, obtaining image texture data and sound spectrum data of a monitored device collected by the cross-domain Internet of Things device; performing cross-modal feature space alignment by using a feature level fusion classifier based on the image texture data and the sound spectrum data, and generating a joint semantic feature vector representing the state of the monitored equipment; constructing a semantic knowledge graph based on the joint semantic feature vector; and dynamically optimizing the feature weight of the cross-domain Internet of Things equipment in the semantic knowledge graph according to the level value, so as to generate a cross-domain Internet of Things equipment cooperative control instruction according to the dynamically optimized feature weight. According to the invention, multi-modal data collaborative sensing and dynamic weight optimization driven by vibration signals are realized, and the monitoring precision and control efficiency of the cross-domain equipment state are improved.

The invention discloses a bar count visual identification automatic calculation system and method, the system comprises a stereo visual acquisition unit, a dynamic threshold adjustment unit, a geometric feature analysis unit, a data processing module and a man-machine interaction unit, and the method comprises the following steps: S01, image acquisition and preprocessing; s02, adjusting an environment self-adaptive threshold value; s03, bar geometric feature extraction; s04, the number of bars is calculated; s05, outputting a result, displaying the bar count obtained by calculation on an interface of a man-machine interaction unit, and storing a counting result and related data in a database at the same time; the beneficial effects of the invention are that a conventional visual counting system usually only depends on two-dimensional image information, recognition errors are easy to occur under the conditions of bar adhesion, surface reflection and the like, the scheme constructs a stereoscopic vision system through an industrial camera to obtain three-dimensional point cloud data, image textures are enhanced in combination with coding structured light projected by a structured light emitter, and the recognition accuracy is improved. And the spatial position and surface depth information can be obtained.

The invention relates to the technical field of computer vision, in particular to an industrial facility autonomous inspection and intelligent diagnosis method based on multi-modal vision fusion, which comprises the following steps of: acquiring a multi-frame seam image to extract a texture dot matrix, analyzing displacement frequency to evaluate stability, calculating angle difference to identify a deformation structure surface, and performing multi-modal vision fusion. And detecting image and point cloud overlapping, screening abnormal blocks to generate a fusion set, identifying a spectrum hopping mapping image, positioning a boundary region, extracting a frequency analysis path repetition rate, identifying an abnormal focusing position, and generating a diagnosis list. According to the invention, through multi-frame image texture dot matrix displacement tracking, the dynamic abnormal region identification precision is improved, micro deformation is captured through three-dimensional point cloud angle difference, the structural anomaly detection is enhanced, the multi-modal anomaly judgment accuracy is improved, the anomaly characteristics, the inspection path frequency and anomaly aggregation analysis are refined, the high-frequency abnormal region is positioned, and the inspection resource configuration is optimized; risk prevention and control are enhanced, and limitation of equipment surface sensing and abnormal positioning is broken through.

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

The invention relates to the technical field of semantic understanding, in particular to a three-dimensional scene semantic understanding method and system based on multi-modal deep learning, and the method comprises the following steps: collecting a point cloud image and a depth map in an automatic driving scene, carrying out the normalization standardization and deletion filling, extracting texture geometric space features, and carrying out the fusion through an attention mechanism; a multi-time-step state vector is introduced to calculate change features, a spatial relation between road participation objects is modeled, a dynamic instance graph structure is constructed, semantic tags are reasoned, and fusion features are compared to generate a three-dimensional scene semantic understanding result. According to the method, the fusion quality is guaranteed through multi-source data normalization standardization, the semantic complementarity is enhanced through collaborative extraction of image texture and point cloud geometric features, the dynamic scene perception ability is improved through state vector modeling, the object interaction semantic relation is described through a spatial relation graph, and the recognition accuracy and consistency are improved through a semantic label reasoning mechanism. And the integrity and robustness of three-dimensional semantic understanding are integrally enhanced.

The invention relates to the technical field of three-dimensional modeling, in particular to a live-action three-dimensional processing method and system based on big data, which are used for extracting height difference mutation and texture features based on remote sensing big data, fusing transparency and direction continuity information, dynamically adjusting a boundary confidence level and realizing weighted reconstruction and three-dimensional expression optimization of a mountain texture layer. According to the method, the elevation grid is established through remote sensing big data, and first-order difference is carried out to identify a height difference mutation region, so that a mountain boundary identification process is changed from dependence on image texture features to quantitative analysis on topographic changes, and boundary state discrimination is carried out in combination with texture layer transparency and gray scale contrast features; according to the method, boundary judgment has continuity and directivity recognition capability, joint evaluation is performed based on a transparency change trend and a boundary direction fitting error, dynamic adjustment of boundary confidence is realized, gray scale reconstruction is performed on a layer boundary region according to a confidence adjustment result, and the accuracy of a layer splicing position and direction is enhanced.

The invention relates to the technical field of agricultural microbial monitoring, in particular to a microbial action monitoring and analysis system which comprises a pollution detection module, a distribution recognition module, a state analysis module, a period extraction module and a node traceability module. According to the method, a pollution identification basis is constructed based on a multi-channel conductance deviation trend and a synchronization feature, flora distribution information is established by introducing a feature grouping mode of image texture, gray level and structural density, and consistency judgment is performed by combining direction deviation of three types of response data of conductance, gas release concentration and pH; the continuity and the staged response precision of microbial function state recognition are enhanced, and meanwhile, a linked list structure index path is constructed according to sampling time, position and channel sequence, so that a continuous mapping process can be formed by a front-back sequence and behavior transition relation of a state period; the structure expression capability and the rhythm construction efficiency of the multivariable monitoring data in the microbial action stage analysis are improved.

The invention discloses a remote sensing image time-space fusion method based on spectrum-spatial feature matching, which comprises the following steps: selecting a remote sensing image data set, obtaining a coarse-fine spatial resolution image pair as a reference image pair, and inputting the reference image pair and a prediction date coarse spatial resolution image into a network to generate a prediction date fine spatial resolution image; according to the network, a spectrum-space matching module (SSFM) is introduced into a GAN-STFM, an expansion convolution branch and a difference-attention module (SUB-AM) are introduced into an encoder, and a DCT module (PyDCT) is added behind an encoder-decoder structure. During training, a definition loss item is added into a loss function of the network. According to the method, the SSFM enables the selected reference image to be more accurate, and the introduced expansion convolution and SUB-AM alleviate the problem of low accuracy of the fused image. The PyDCT and the definition loss item solve the problems that the texture detail part of the fused image is missing and the region edge is fuzzy.

The invention provides a liver and gall disease prediction model construction method, system and device combined with multi-modal data and a medium, and belongs to the technical field of data prediction model construction. Medical data of a patient is collected; extracting the medical data in the medical data set, and evaluating the feature weight of each piece of medical data by using a regression analysis statistical algorithm; a comprehensive feature vector set is generated through fusion, and a preliminary liver and gall disease prediction model is constructed; and deploying the liver and gall disease prediction model to a medical terminal, and performing periodic updating through a cloud. A regression analysis statistical algorithm is utilized to evaluate a feature weight, features, such as key features such as glutamic-pyruvic transaminase and liver ultrasound image texture features, which have important influences on liver and gall disease prediction can be screened out, the data dimension is reduced, the model complexity is reduced, the key feature effect is highlighted, and the interpretability and prediction accuracy of the model are improved.

The invention relates to the technical field of three-dimensional modeling, and discloses a hybrid modeling method, device and equipment, and the method comprises the steps: constructing a three-dimensional model for any part of a target structure in modeling software, the three-dimensional model being a grid model or a curved surface model; fusing the three-dimensional models of all the components under geometric constraints to obtain a first mixed model; flattening the first mixed model in a two-dimensional plane to obtain a two-dimensional coordinate; superposing image textures on the two-dimensional coordinates and performing data compression to obtain texture coordinates; mapping the texture coordinates to a three-dimensional space to obtain a second mixed model; and individually setting the second hybrid model, and exporting a third hybrid model of the target structure through rendering. According to the method, a systematic mixed modeling process is formed by integrating the steps of geometric modeling, seamless fusion, flattening, texture superposition, rendering and the like, the working efficiency is improved, it is ensured that the finally output model has high precision and high visual effect, and meanwhile consumption of storage and computing resources is reduced.

The invention relates to a low-illumination image enhancement method based on multi-mode classification and brightness feedback, and belongs to the field of image processing. The method comprises the following steps of: firstly, training a network, constructing a multi-modal illumination prior feature of an image, inputting the multi-modal feature and a brightness score of the image into an illumination perception classification network, obtaining a local probability value of image brightness, adaptively selecting a local or global enhancement processing method, and generating a fusion weight; in the local enhancement processing, dark area details are enhanced through a multi-scale parallel and double-attention mechanism, in the global enhancement, the brightness is improved by using a symmetric coding-decoding structure and a residual attention block, and linear fusion is performed after the brightness is enhanced; then, the brightness evaluation value of the image is fed back through the lightweight brightness estimation network, and the brightness of the image is enhanced. And optimizing the training network according to the joint loss function of image processing. The method provided by the invention can enhance image texture details and improve image definition under a non-uniform illumination condition and an extremely low illumination condition.

The invention provides an ovarian cancer accurate diagnosis and personalized treatment method based on data integration, and the method comprises the steps: employing a feature extraction algorithm to generate a multi-dimensional feature vector set containing a gene expression level, an image texture parameter, a symptom score and a pathological classification according to a structured data set; performing dimension reduction processing on the multi-dimensional feature vector set through a feature fusion module to obtain a low-dimensional feature representation vector; performing grouping and risk prediction on the patient data by adopting a classification algorithm according to the low-dimensional feature representation vector, and generating a patient subgroup classification and disease risk layering result; according to the patient subgroup classification and the disease risk layering result, a scoring model is adopted to generate a diagnosis scoring result and a personalized treatment recommendation scheme, and a diagnosis report is output.

The invention discloses a deep sea mineral resource segmentation method and device based on dynamic anchor points and iterative optimization, and relates to the field of computer vision, and the method comprises the steps: obtaining sonar point clouds and laser radar point clouds of deep sea mineral resources, combining the sonar point clouds and the laser radar point clouds into unified point clouds, and obtaining high-definition image texture features; fusing the initial geometric features and the texture features of the unified point cloud to obtain initial multi-modal features; constructing a kernel point, calculating a local structure feature and obtaining a multi-scale feature; using a graph attention network to extract global features and performing clustering to obtain class anchor points; fusing the preliminary multi-modal features, the texture features and the global features to obtain a final feature representation, calculating the similarity between the final feature representation and the class anchor points, and performing classification to obtain a preliminary segmentation result; boundary smoothing and texture correction are carried out on the preliminary segmentation result, and then fusion with multi-scale features is carried out; suspicious points are detected, error correction is carried out on the suspicious points, and segmentation is completed. According to the method, multi-modal feature iterative optimization, misclassification point continuous correction and class anchor point timely updating are realized.

The present application discloses a video encoding method and apparatus, a device, and a readable storage medium. The method comprises: acquiring the current encoding bit rate of a video encoder; acquiring a frame sequence to be encoded of a target video source, and starting to encode an initial video frame in said frame sequence by means of the video encoder on the basis of the encoding bit rate; determining a first image texture value of a first video frame currently prepared to be encoded in said frame sequence, and determining a second image texture value of a second video frame immediately preceding the first video frame, the first video frame being any video frame following the initial video frame; when the first image texture value is greater than the second image texture value, determining an image texture ratio on the basis of the first image texture value and the second image texture value; and adjusting the encoding bit rate of the video encoder on the basis of the image texture ratio to obtain a target encoding bit rate, and encoding the first video frame on the basis of the target encoding bit rate. In this way, the bit rate of the video encoder is adjusted at the frame level, ensuring the quality of encoded video frames.

The invention relates to the technical field of image modeling, in particular to a three-dimensional modeling method and system for medical image data, and the method comprises the steps: carrying out the preprocessing of the obtained two-dimensional or three-dimensional medical image data, extracting a blood vessel region based on the image texture and gray features, and generating a preliminary blood vessel structure segmentation image; in combination with a blood flow simulation or real acquisition means, obtaining dynamic feature data from the vascular structure segmentation map, and mapping the dynamic feature data to a corresponding vascular structure position to form a vascular distribution map with blood flow constraint; and inputting the blood vessel distribution diagram into a three-dimensional modeling module, expanding and complementing small blood vessel branches by utilizing blood flow constraint, dynamically correcting the form of a main blood vessel, and generating a three-dimensional blood vessel structure model with real fluid physiological characteristics. The method can be widely applied to medical research and clinical aid decision-making scenes such as preoperative simulation, individualized intervention path planning and blood flow reconstruction analysis, and has good popularization and application prospects and engineering practical value.