475 results about "Radiation field" patented technology

The invention discloses a dynamic semantic SLAM (Simultaneous Localization and Mapping)-driven robot full life cycle navigation method, which comprises the following steps: synchronously acquiring multi-modal data through a multi-modal sensor array, and carrying out joint coding based on a cross-modal contrast learning framework to obtain embedded features. A dynamic three-dimensional Gaussian radiation field is constructed, a scene is represented as a parameterized Gaussian kernel set, and mixed Gaussian field representation is obtained through micro rendering and joint training optimization. A Gaussian mixture field is divided into a global static layer and a local dynamic layer, retention constraints are applied to the static layer, and an incremental online updating strategy is adopted for the dynamic layer. Based on the optimized Gaussian field parameters and the multi-modal data, a factor graph containing Gaussian rendering residual factors and inertia pre-integration factors is constructed and solved, the corrected robot pose is obtained, and the space occupancy probability is generated. Operation control parameters are adjusted on line according to the pose and the occupancy probability through a strategy network, and autonomous navigation is achieved in combination with path planning and trajectory tracking.

The invention discloses a deep foundation pit crack feature coding detection method based on image recognition, and aims to solve the problems that a two-dimensional image is difficult to accurately characterize a three-dimensional crack, and the opening width and the slab staggering height are inaccurately estimated, the deep foundation pit crack feature coding detection method comprises the following steps of: calibrating and collecting through short time sequence and multi-polarization imaging, segmenting the crack to extract a center line, and dividing a left side domain and a right side domain; carrying out local reconstruction in a mask neighborhood by adopting a nerve radiation field or three-dimensional Gaussian sputtering, establishing a left implicit surface and a right implicit surface, fusing a polarization solution line, applying occupancy priori and seam-crossing repulsive potential in a gap region, combining a reference object to obtain an absolute scale, measuring a millimeter-level opening and a dislocation along a central line section, and generating a feature code; and threshold value and time sequence comparison judgment is carried out, and the technical effects of high-precision three-dimensional measurement, reliable risk assessment and traceable management of the deep foundation pit crack are achieved.

The invention discloses a radiation safety management method and system based on cloud platform data driving, and relates to the technical field of cloud platform radiation management. The method comprises the following steps: collecting radiation field data and environment state data of a target area through a deployed intelligent sensing node; uploading the radiation field data, the environment state data and the context data from the service module to a cloud platform, performing energy compensation and radiation unmixing, and constructing a multi-dimensional feature vector; and processing the multi-dimensional feature vector by using a pre-trained situational radiation perception model, identifying radiation field features and an environment situation, and generating a situational radiation safety early warning signal. The technical problem that in the prior art, radiation safety monitoring depends on single-point measurement, comprehensive judgment cannot be carried out in combination with environment and service context data, and consequently the radiation field anomaly recognition capability is insufficient is solved, and the purpose that the radiation field anomaly recognition capability is improved through cloud platform data driving and context awareness model fusion is achieved. And the technical effects of high-precision identification and situational safety early warning of the radiation field state are realized.

The invention discloses a three-dimensional radiation field construction method and system, and the method comprises the steps: firstly obtaining an actual three-dimensional coordinate parameter set and an actual radiation dose rate parameter set of a to-be-measured scene, constructing a standard covariance matrix according to the parameters, selecting a to-be-predicted point, generating an actual covariance vector for the to-be-predicted point, and carrying out the calculation of the actual covariance vector. Generating predicted radiation dose rate parameters based on the standard covariance matrix and the actual covariance vector, and collecting the predicted radiation dose rate parameters of the plurality of points to be predicted into a predicted radiation dose rate parameter set, and finally, generating a three-dimensional radiation field according to the actual three-dimensional coordinate parameter set, the actual radiation dose rate parameter set and the predicted radiation dose rate parameter set. According to the method, the correlation between the trajectory points is represented through the covariance, and the conversion relation between the to-be-predicted point and each trajectory point is established through the conversion relation between the actual covariance vector and the standard covariance matrix, so that the radiation dose rate of the predicted point is calculated, the radiation dose rate of more point positions is obtained, and the precision of the three-dimensional radiation field is improved.

The invention provides a two-stage human body posture estimation model lightweight method and system based on pseudo-view data generation and a storage medium. The method comprises the following steps: step 1, in a first stage, performing cross-model knowledge migration based on a multi-scale feature map and a regression logic value; step 2, a data enhancement strategy generated based on a middle visual angle: utilizing existing multi-visual-angle image data, through implicit three-dimensional representation of a neural radiation field learning scene, synthesizing a pseudo-visual-angle image between existing camera visual angles and a corresponding attitude pseudo label, mixing the generated pseudo-visual-angle data with a real data set, and obtaining a data enhancement strategy; commonly used for training a student model; and step 3, a second stage, self-distillation refining and anatomical optimization. The method has the beneficial effects that the knowledge of the teacher model is more finely transmitted to the student model through a structured knowledge migration process, and effective light weight of the model is realized on the premise of ensuring the accuracy and robustness of attitude estimation, so that the model meets the requirements of actual application on efficiency and deployment.

The invention discloses a sparse visual angle three-dimensional reconstruction method and system based on voxel grid constraint, and belongs to the technical field of visual three-dimensional reconstruction. Constructing a three-dimensional voxel grid of a self-adaptive scene scale based on point cloud distribution, and dividing the point cloud to the corresponding voxel grid; fusing the geometric features of the fast point feature histogram and the confidence score in the grid, generating a geometric confidence comprehensive measure, and screening key points to initialize Gaussian primitives; designing a voxel grid constrained gradient clipping strategy, limiting Gaussian primitive error diffusion through a distance attenuation coefficient, and adaptively optimizing grid distribution in combination with a dynamic grid deletion and addition mechanism; and finally, carrying out iterative training by using a 3D Gaussian splash radiation field loss function to realize high-fidelity static scene reconstruction under a sparse view angle. According to the method, scene geometric priori is introduced, an optimization strategy based on voxel grid constraint is designed to effectively control excessive diffusion or drift of Gaussian primitives, generation of artifacts is reduced, and meanwhile, the situation that robustness is reduced due to the influence of priori quality is avoided.

The invention discloses a three-dimensional packaging modeling method based on NeRF and a diffusion model, and the method comprises the steps: firstly receiving demand description information, fusing packaging physical constraint conditions for joint coding, and forming a comprehensive condition vector; secondly, a cross-path dynamic information exchange mechanism is established through a geometric generation path and a texture optimization path, an initial three-dimensional packaging model is generated, physical consistency constraints are added, and NeRF is used for conducting volume rendering reconstruction on the initial three-dimensional packaging model; and then introducing an observation distance adaptive dynamic scale perception neural radiation field optimization mechanism, and generating a multi-scale NeRF model after scale perception optimization. And finally, establishing mapping of physical attributes and optical characteristics of the packaging materials, and generating a final three-dimensional packaging model through independent optical calculation and physical synthesis processes of the multi-layer materials. According to the method, the semantic understanding deviation of CLIP coding in vertical field application is solved, and the material reality representation of industrial-grade three-dimensional packaging modeling is realized.

The invention provides an electron irradiation dose on-line measurement and feedback calibration system and method, and relates to the technical field of automation control, and the system comprises a processing module which is used for converting an original dose signal into a five-path digital reading sequence, carrying out the timestamp alignment and space coordinate calibration of the five-path digital reading sequence, and outputting a calibration result; generating a three-dimensional discrete coordinate data set representing radiation field intensity space distribution; and the extraction module is used for identifying a radiation intensity global maximum value point from the three-dimensional discrete coordinate data set as a spatial registration origin, extracting two feature straight lines representing a radiation scattering dominant direction, and calculating a spatial orientation included angle between the two feature straight lines to determine a radiation scattering main influence area. According to the invention, precise processing of an original dose signal, scientific positioning of a radiation influence area and construction of a dynamic compensation model are realized through cooperation of multiple modules, a high-confidence dose value can be output, feedback calibration is completed based on a time sequence drift trend, and the precision and reliability of on-line measurement of the electron radiation dose are effectively guaranteed.

The invention provides a factory three-dimensional digital twin model reconstruction and incremental updating method for industrial space intelligence, and the method comprises the steps: obtaining a factory multi-view basic image for a factory panoramic scene, so as to train a bifurcated structure neural implicit field network containing a radiation field branch and an SDF branch, and generating an initial three-dimensional digital twin base; acquiring factory real-time inspection images acquired by the inspection robot under different inspection poses, rendering the initial three-dimensional digital twin base by using a radiation field branch to obtain a virtual reference view, and performing difference analysis on the virtual reference view and the factory real-time inspection images to position a structural change area in a factory; determining an axis alignment bounding box for changing the physical range of the entity in the area, taking the axis alignment bounding box as an effective boundary of the local space patch, and determining matched radiation field network weight and SDF network weight according to the effective boundary; and adjusting the radiation field network weight and the SDF network weight, and updating the radiation field network weight and the SDF network weight to the network in an incremental updating manner to generate a factory three-dimensional digital twinning model based on the implicit three-dimensional geometric field.

The invention relates to the technical field of borehole fracture analysis, discloses a borehole fracture evaluation method and system based on a neural radiation field and acoustic emission, and aims to solve the problem of poor accuracy in the prior art. The scheme mainly comprises the following steps: constructing a high-fidelity hole wall three-dimensional model through a neural radiation field technology; meanwhile, time-space coordinates and energy parameters of an acoustic emission event are obtained through an acoustic emission sensor array; uniformly converting acoustic emission coordinates into a model coordinate system, and identifying and quantifying geometric features of the fissure surface; and finally, constructing a space-time diagram structure taking the fracture surface as a static node and the acoustic emission event as a dynamic node, learning a space-time evolution mode through a graph neural network, outputting a fracture surface activity index, and realizing activity grade evaluation and extension trend prediction. The method improves the accuracy of borehole fracture analysis, and is particularly suitable for safety monitoring of geotechnical engineering.

The invention discloses a dynamic calculation method and system for a radiation field in the cutting process of an activity non-uniform radioactive source, and belongs to the field of radiation field calculation. The method comprises the following steps: acquiring three-dimensional activity distribution data of a radioactive source; constructing a voxelization model and coupling the voxelization model with an activity grid to distribute activity; establishing an LBVH hierarchical structure of the shielding body; a point kernel is generated in a self-adaptive mode through ray tracing and average free path comparison; carrying out radiation field calculation by adopting point nuclear integration, and realizing multilayer ray tracing in the radioactive source, between the sources and a shield in the process; on-line cutting and moving operation of the voxel model is supported, and the radiation field is dynamically updated. According to the method, the cutting operation of the activity non-uniform radioactive source can be simulated, the problem that the radioactive activity of each cutting product cannot be represented after cutting by a traditional method is solved, the radiation field dynamic calculation in the activity non-uniform radioactive source cutting process can be simulated, initial modeling of the radiation environment only needs to be carried out once, and the efficiency is improved. And repeated modeling is not needed.

The invention belongs to the technical field of artificial intelligence and three-dimensional scene reconstruction, and discloses a sparse sampling three-dimensional scene reconstruction method based on cognitive priori guidance. Comprising the following steps: data preprocessing and scene structure recovery; initializing a radiation field; extracting cognitive embedding and view priori generation; monocular depth constraint is carried out; reconstructing a three-dimensional scene; and calculating a loss function. According to the method, global cognition of a to-be-reconstructed scene is fully learned by constructing a scene feature cognition encoder, a pre-training generation model is guided to provide an effective training supplement view, geometric consistency of a radiation field is constrained by adopting depth information, distribution of Gaussian primitives is adjusted to solve the problem that a scene geometric structure is learned mistakenly, and the scene reconstruction efficiency is improved. And the quality of three-dimensional scene reconstruction in a sparse sampling scene is greatly improved. By using the method, scene global cognitive embedding can be effectively established, meanwhile, the geometric consistency of the scene is restrained, a compact and effective Gaussian radiation field is constructed, and the three-dimensional scene reconstruction quality under the sparse sampling condition is remarkably improved.

The invention discloses a rapid large-scale calculation simulation method and system for gamma radiation dose in a dynamic environment and a storage medium, and belongs to the technical field of ionizing radiation protection. The method comprises the following steps: firstly, constructing a triangular mesh scene model containing a radioactive source and a shielding body, and discretizing the radioactive source into a local point kernel; global coordinates of the point kernel and the vertex of the triangle are calculated in parallel by using a GPU, and a maximum parallelization hierarchical surrounding structure MPBVH of the scene is constructed; ray tracing is executed in parallel on the GPU, and all intersection points of MPBVH calculation rays and the shielding body are traversed to determine the thickness and the material of each shielding layer; and finally, dose values of all counting grids are calculated in parallel by adopting a point kernel method. When a radioactive source or a shielding body in a scene is subjected to dynamic changes such as movement and rotation, geometric data can be quickly updated, and calculation can be executed again. According to the method, the speed of large-scale point kernel calculation is increased, the influence of the environmental complexity on large-scale calculation time consumption is small, and the method can be suitable for rapid large-scale calculation and dynamic updating of the radiation field in the complex dynamic environment.

The invention discloses a building fire safety management and control method based on a digital twinborn technology, and the method comprises the steps: constructing sensor-middle platform-model bidirectional mapping, and reconstructing a dynamic scene through employing a neural radiation field technology; fusing the cross-domain desensitization data based on federal transfer learning, and constructing a building total-factor dynamic digital mirror image; through a spectrum analysis algorithm, a fire source is positioned in real time, and a flame spreading track is predicted; performing fire simulation in linkage with a total-factor dynamic digital mirror image, and optimizing a fire-fighting resource scheduling path through a multi-target particle swarm optimization algorithm; a multi-target rescue scheme is generated, an asynchronous block chain is used for evidence storage of a fire-fighting decision instruction, a rescue key area is plotted, and an optimal evacuation route is planned; and converging federal node data to quantify the data contribution degree, reproducing a historical fire behavior and deducing a new scene to carry out iterative optimization. According to the method, the problems of data sensing fragmentation, difficulty in cross-domain fusion, fire prediction lag, underscientific resource scheduling, slow drill scheme iteration and the like in traditional management and control are solved.

The invention discloses a convolutional neural network and multi-source data fusion-based nuclear radiation on-line monitoring method, which comprises the following steps of: acquiring laser radar point cloud data, a radiation dose rate and a radiation image of a monitoring area by adopting an unmanned aerial vehicle, and de-noising the radiation image by adopting a double-branch convolutional neural network; determining the track position of the unmanned aerial vehicle according to the laser radar point cloud and the de-noised image; performing space-time registration on the radiation dose rate and the track position of the unmanned aerial vehicle to form a data set, and obtaining a continuous radiation field of the monitoring area by adopting Kriging interpolation; according to the continuous radiation field, an L-BFGS-B algorithm is adopted to carry out minimization solution on a loss function with the radiation source position as an optimization variable, and position coordinates of the radiation source are obtained; performing data cleaning on the radiation dose in the continuous radiation field, and then performing multi-dimensional feature extraction to obtain a feature vector; and inputting the feature vectors into a plurality of prediction models to obtain a plurality of predicted radiation doses, and carrying out weighted fusion to obtain a predicted dose rate of the radiation source.

The invention discloses a probabilistic thermal runaway early warning method and system based on a multi-modal three-dimensional thermal field, and relates to the technical field of energy storage safety monitoring, and the early warning method comprises the specific steps: S100, multi-modal data collection; s200, carrying out time alignment processing; s300, feature fusion extraction is carried out; s400, reconstructing a three-dimensional temperature field; s500, calculating a thermal runaway probability; according to the method, multi-modal data of RGB-D, long-wave infrared and gas sensors are fused, space-time synchronization is realized in combination with a soft Kalman alignment algorithm, the limitation of single physical quantity monitoring is broken through, cross-modal correlation features are extracted by adopting a pyramid cross attention adversarial network, and the multi-modal data of the RGB-D, the long-wave infrared and the gas sensors are subjected to hierarchical linkage response. The generated multi-scale feature tensor can accurately describe the spatio-temporal evolution law of a thermal runaway precursor, after lightweight neural radiation field voxel reconstruction, millimeter-level dynamic updating of a three-dimensional temperature field is realized, the generation and diffusion process of local hot spots in an energy storage cabin can be captured in real time, and a high-fidelity thermal field portrait is provided for early risk identification.

The invention relates to the technical field of radiation environment monitoring, and discloses a dose calculation method and system based on radiation environment monitoring data. The method comprises the following steps: acquiring radiation source data, calculating initial radiation field distribution based on the data, obtaining an initial dose deposition value of each space point in a monitoring area, analyzing interaction among particles, and determining an energy redistribution coefficient of each space point; adjusting the initial dose deposition value point by point by using the coefficient to obtain corrected three-dimensional dose concentration field data, calculating a gradient vector between spatial points, and generating an optimized gradient distribution diagram; extracting a boundary point set with gradient change exceeding a threshold value, constructing a contour surface grid to form a three-dimensional contour surface structure, obtaining a time sequence radiation source data iteration updating coefficient, and obtaining a concentration field time sequence evolution sequence; the high-concentration area is identified and marked based on the sequence, the high-concentration area and the three-dimensional contour surface structure are fused, a visual distribution model reflecting the dose concentration is generated, and the accuracy of radiation dose calculation is improved.

The invention discloses a garden disease and pest identification method and system based on computer image processing, and belongs to the technical field of disease and pest monitoring, and the method specifically comprises the steps: collecting a visible light image, an infrared image and environment parameters of vegetation; performing space-time registration on the visible light image and the infrared image and generating a three-dimensional temperature field model; the three-dimensional space is divided into grid units, and the temperature statistical characteristics and the vertical direction temperature gradient of each unit are extracted; obtaining an expected temperature gradient direction through a preset standard three-dimensional radiation field reference model; and when the actual gradient direction is consistent with the expected gradient direction, calculating a gradient direction deviation value of each grid unit, calibrating the grid unit with the deviation exceeding a threshold value as an abnormal region, and extracting a corresponding visible light image to carry out pest recognition. According to the invention, early-stage accurate identification of diseases and insect pests is realized, systematic diseases and local insect pests are effectively distinguished, interference of environmental factors is overcome, and the reliability of insect pest monitoring is improved.

The invention relates to the technical field of meteorological observation, in particular to a solar radiation prediction method and device based on multi-source satellite data fusion, an electronic device and a storage medium. And an end-to-end deep learning prediction model is constructed, so that a dark start prediction blind area caused by visible light deficiency before and after sunrise is effectively overcome, continuous prediction in all days, especially in a dawn period, is realized, and the prediction coverage capability is expanded. And secondly, regional adaptability parameter calibration is carried out on the physical inversion model, so that the localization precision of a historical reference radiation field is improved, and a more reliable training basis is provided for the prediction model. Finally, the output solar radiation prediction product with high temporal-spatial resolution and physical consistency can directly support key businesses such as photovoltaic power station output prediction and power grid climbing management, and the practicability and reliability of the prediction result are enhanced.

The invention discloses a drainage pipeline three-dimensional imaging method and system based on neural radiation field and Gaussian sputtering fusion reconstruction, and relates to the technical field of computer vision and deep learning. The method comprises the following steps: performing calculation training on an acquired standardized multi-view image to generate point cloud image data, performing initialization sparse three-dimensional structure reconstruction by adopting Gaussian sputtering, guiding neural radiation field initialization by utilizing a sparse three-dimensional structure, generating a rendering image for the same view angle through a Gaussian sputtering model and a neural radiation field network respectively, and performing point cloud image reconstruction. Fusing the two rendering images to obtain a three-dimensional imaging model; according to the method, efficient, high-precision and physically credible three-dimensional reconstruction under the pipeline internal environment with uneven illumination, serious shielding and sparse texture is realized through technology fusion and domain knowledge embedding, and a reliable data basis is provided for pipeline disease diagnosis and quantitative analysis.

The invention belongs to the technical field of automatic control, and particularly relates to a large-span ellipsoidal glass structure space noise reduction method based on sound field optimization, and the method comprises the steps: 1, transmitting a sample noise in an ellipsoidal glass structure through a sound wave generator, building a sound wave propagation model to describe the behavior of the sample noise in the ellipsoidal structure, an incident sound field is obtained; 2, analyzing the propagation characteristics of the sample noise in the ellipsoidal glass structure to obtain an acoustic structure coupling equation, and solving the acoustic structure coupling equation to obtain structural vibration displacement; 3, obtaining a sound radiation field of the ellipsoidal glass structure according to the structural vibration displacement; and step 4, combining the incident sound field and the sound radiation field, and performing spatial noise reduction by reflecting the sound pressure of the sound field. According to the method, the problems of poor adaptability and low control precision of a traditional noise reduction method in a non-uniform structure are effectively solved, and the noise reduction efficiency, the control sensitivity and the environment adaptability of a building space are remarkably improved.

An MLC is provided, the MLC includes: a plurality of leaves; first driving mechanisms connected to the leaves and configured to drive the leaves to move to form a radiation field; and at least one leaf head mechanism mounted on at least one corresponding leaf of the leaves and configured to adjust the radiation field.

The invention discloses a phase-free antenna measurement method, a phase-free antenna measurement system and a storage medium. The phase-free antenna measurement method comprises the following steps: constructing a measurement system comprising an antenna to be measured, a measurement probe, a fixed auxiliary reference antenna and software defined radio; the probe is controlled to move and sample on the scanning plane, and the SDR synchronously collects receiving signals of the probe and the reference antenna; based on the frequency domain transformation result of the two paths of signals, the radiation field amplitude of each sampling point and the relative phase relative to the reference antenna are directly calculated through non-iterative operation, so that complete complex electric field distribution on the scanning plane is reconstructed; and finally, obtaining a far-field directional diagram through near-field-far-field transformation. The fixed reference antenna is introduced to serve as a phase reference, synchronous acquisition and direct calculation are achieved in combination with the SDR, the antenna radiation field can be rapidly and accurately recovered without direct phase measurement or complex iteration in the single scanning plane and single scanning process, and the method has the advantages of being simple in system, high in measurement efficiency and high in signal drift resistance.

The invention relates to the technical field of electromagnetic wave detection and imaging, and discloses a radar target echo generation method based on a physical information neural radiation field, which effectively improves the physical consistency of generated echo signals, accurately simulates the energy attenuation characteristics at different observation angles, and improves the radar target echo generation efficiency. The problem of signal completion under the sparse view angle condition is solved. The signal generated by the method is highly matched with measured data in the aspects of amplitude distribution, phase continuity and energy envelope form, a reliable data basis is provided for radar target identification and feature analysis, and the perception capability of a network on target geometric details is enhanced through high-dimensional position coding. Therefore, the generated echo signal can accurately reflect the spatial distribution characteristics of the scattering center. Meanwhile, due to the introduction of the multi-frequency coding sequence, the network can adaptively capture electromagnetic scattering characteristics under different scales, and a reliable technical means is provided for echo generation of a complex target.

The invention discloses a composite shielding collimator suitable for a complex mixed radiation field, belongs to the technical field of radiation shielding and measurement, and mainly solves the problems of low signal-to-noise ratio, high space background and mixed energy spectrum caused by the mixed radiation field of high-energy neutrons, gamma rays and the like in a fusion device and the problem that traditional equipment is heavy and difficult to flexibly adjust. The composite shielding collimator adopts a cylindrical shielding structure formed by aluminum oxide, boron-containing polyethylene, boron carbide and lead from outside to inside, and background interference and secondary radiation are remarkably inhibited by moderating and absorbing neutrons step by step and effectively shielding gamma rays; the neutron and gamma ray detector is integrally integrated on a rotatable supporting mechanism, unification of excellent radiation shielding performance and detection flexibility is achieved, and the neutron and gamma ray detector is mainly used for high-precision neutron and gamma ray diagnosis in complex radiation environments such as a magnetic confinement fusion device.

The invention relates to the technical field of rice phenotype extraction, and discloses a rice fine phenotype extraction method based on three-dimensional vision, a terminal and a storage medium. The method comprises the following steps: acquiring a multi-view image of a rice plant by using a depth camera, and performing camera attitude estimation on the multi-view enhanced image to obtain camera internal and external parameters corresponding to each view; constructing a neural radiation field network, training the neural radiation field network by taking the enhanced image, the contour mask and internal and external parameters of the camera as input, and learning color and density distribution of a scene by utilizing a volume rendering technology; extracting surface points based on a mixed threshold strategy by using the trained neural radiation field network, and generating a rice three-dimensional point cloud; inputting the rice three-dimensional point cloud into a point cloud instance segmentation network, and outputting an instance mask containing each organ of the rice and a segmentation result of a three-dimensional bounding box; and counting phenotypic parameters of the rice according to a segmentation result. According to the method, the fine instance segmentation of the high-density point cloud is realized while the reconstruction precision is ensured, and the phenotypic analysis of the spike fraction is supported.

The invention relates to the technical field of image processing, and discloses a three-dimensional reconstruction method and system based on BARF and ZoeDepth, and the method comprises the following steps: S1, collecting a multi-view RGB image of a target scene; s2, taking each acquired RGB image as an input image, performing single image depth estimation through a pre-trained ZoeDepth model, and generating a corresponding depth map; s3, taking the generated depth map as geometric priori, and integrating the depth map into a BARF framework to obtain an optimized neural radiation field model; and S4, dense point clouds are extracted from the optimized neural radiation field, and a grid model is generated through a surface reconstruction algorithm. According to the method, under the condition that the number of input images is seriously limited, strong geometric constraints provided by depth information are fully utilized, a high-precision and high-integrity three-dimensional model is quickly and stably reconstructed, and the method has good engineering practical value.

The invention discloses a metal product surface defect detection method and system based on machine vision, and the method comprises the steps: synchronously collecting three-dimensional geometric and two-dimensional texture information through structured light and a multispectral composite sensor, and obtaining clear initial defect data through a reflection separation technology based on the degree of polarization; precise positioning of sub-pixel-level defect boundaries is realized by using multi-modal data alignment and a contour enhancement technology based on an attention mechanism, multi-task intelligent classification is further performed by fusing contour, geometric and spectral features, and fine distinguishing and quantitative description of defects are completed. A high-fidelity three-dimensional digital model of key defects is constructed through a neural radiation field technology, model self-optimization is realized in combination with view difference and generative completion, multi-dimensional quantitative indexes and a process knowledge rule base are subjected to automatic matching reasoning, a structured detection report is generated, full-process intelligence from perception, analysis and reconstruction to decision making is realized, and the detection efficiency is improved. And the accuracy, efficiency and automation level of industrial quality control are obviously improved.

A computer-implemented method includes encoding a radiance field of an object onto a machine learning model; conducting, based on a set of training images of the object, a training process on the machine learning model to obtain a trained machine learning model, wherein the training process includes a first training process using a plurality of first test sample points followed by a second training process using a plurality of second test sample points located within a threshold distance from a surface region of the object; obtaining target view parameters indicating a view direction of the object; obtaining a plurality of rays associated with a target image of the object; obtaining render sample points on the plurality of rays associated with the target image; and rendering, by inputting the render sample points to the trained machine learning model, colors associated with the pixels of the target image.

The invention discloses a three-dimensional indoor scene generation method and related equipment. The method comprises the steps that text description of an indoor scene is acquired; generating an indoor layout by the text description through a preset diffusion model; converting the indoor layout into a two-dimensional height field and a two-dimensional semantic graph; predicting a neural radiation field based on the two-dimensional height field and the two-dimensional semantic map; a plurality of viewpoints are sampled in the predicted neural radiation field, the plurality of viewpoints are rendered to obtain an RGB-D image corresponding to each viewpoint, RGB is color information of the two-dimensional image, and D is depth information of the two-dimensional image; and according to the RGB-D image corresponding to each viewpoint, constructing a truncated signed distance field body, and generating a mesh triangular patch based on the truncated signed distance field body to obtain an indoor three-dimensional scene. According to the method and the device, the high-quality three-dimensional indoor scene can be generated without real multi-view images, the efficiency of generating the three-dimensional scene is improved, and the method and the device can be widely applied to the technical field of computer vision.