489 results about "Rasterisation" patented technology

The invention discloses a Gaussian representation SLAM method based on dense matching priori and factor graph constraint, which comprises the steps of inputting a current image and a key frame image, outputting a point graph corresponding to the image through a pre-trained model, returning a matching condition of two frame image points and respective point cloud information, and obtaining a point-level matching result based on a point-level matching result. The method comprises the following steps: constructing a joint optimization problem of a current frame and a key frame by taking a luminosity consistency error and a geometric projection error as targets, performing joint estimation on a camera pose and a point cloud of the current frame, realizing high-precision pose solution, generating point diagram data after Gaussian scene representation and rasterized rendering processing, and transmitting the point diagram data to a rear end for global optimization. And the rear end receives the pose and point cloud data, executes loopback detection to identify repeated key frames, and performs Gaussian rendering through an optimized key frame image to complete global dense three-dimensional reconstruction. The method effectively solves the problem of track drift and scene inconsistency caused by lack of pose priori and global geometric constraints in an existing system.

The invention provides a lake water quality monitoring method and system based on deep learning, and relates to the field of deep learning. The method comprises the following steps: firstly, collecting original sensor data, remote sensing images, hydrodynamic force and meteorological data, carrying out space-time alignment processing, and constructing a lake surface rasterization space-time diagram; the graph is input into a model composed of a spectrum attention encoder, a graph convolutional network and a time sequence unit, lake global water quality index concentration distribution is predicted, a dynamic threshold value is set based on historical data, the standard exceeding condition is recognized in real time, pollution early warning is triggered, and high-precision monitoring and intelligent management of lake water quality are achieved.

The invention relates to the technical field of unmanned aerial vehicle cooperative tracking, in particular to a multi-unmanned aerial vehicle cooperative tracking method and system based on a complex environment. Comprising the following steps: S1, acquiring image data acquired by each unmanned aerial vehicle and flight position data of each unmanned aerial vehicle, and determining a tracking target, a main tracking unmanned aerial vehicle and an auxiliary unmanned aerial vehicle; s2, establishing an appearance feature library for the tracking target according to the image data; an auxiliary unmanned aerial vehicle checks an environment shielding risk in advance through rasterizing a pre-survey point and simulating a target moving track, when shielding occurs, a stop mechanism is triggered, an auxiliary tracking position is optimized, an optimal position is calculated in combination with a direction weight and a distance threshold value, continuity of a tracking process is ensured, and the tracking efficiency is improved. Meanwhile, when the image feature proportion of the auxiliary unmanned aerial vehicle is higher than that of the main tracking machine, identity replacement is automatically triggered, it is ensured that when the view angle of the main tracking machine is limited, continuous recording of the target can still be kept through the auxiliary unmanned aerial vehicle, and the task reliability in the complex environment is improved.

The invention discloses an island reef shoreline remote sensing extraction method based on a double-branch network, and belongs to the technical field of island reef shoreline remote sensing extraction. Comprising the following steps: S1, acquiring an original satellite remote sensing image of a coral island reef, and cutting to obtain a reef remote sensing image; s2, a shoreline vector label is drawn in the island remote sensing image, and rasterization processing is carried out; s3, cutting the rasterized label and the research area image to obtain tile data; s4, dividing the tile data into a training set and a test set; s5, performing data enhancement on the training set tile data to obtain a processed image data set; and step S6, constructing an SDSPNet model based on the image data set. According to the method, the overall semantic expression ability of the broken shoreline and the lagoon face area is enhanced, and the problems of erroneous judgment of the broken shoreline of the coral island and fuzzy lagoon face boundary in an existing method are solved.

The invention discloses an urban street land fine division method and system based on deep learning, and belongs to the field of geographic information processing. The method comprises the following steps: firstly, collecting remote sensing image data, building space form data, business form POI data and urban land data of a case city district, extracting various kinds of information, converting the information into two-dimensional plane image format data, and carrying out rasterization and data enhancement processing; then training the land use division convolutional neural network deep learning model through a grid translation window method and a grid land use function judgment rule, performing rasterization function judgment on urban land use in a target area, dividing grid boundaries through a watershed method, and finally checking and correcting a result to obtain a target urban land use function judgment result. And obtaining a final result of land division of the target district. According to the method, the internal land of the urban land street can be finely divided, automatic division and function judgment of the urban land are realized, and digital and intelligent application conversion of urban planning work is promoted.

The invention belongs to the technical field of navigation, and relates to an unmanned aerial vehicle multi-target image acquisition path planning method, equipment and a storage medium. The method comprises the following steps: acquiring a city live-action three-dimensional model; rasterizing the building model to create a three-dimensional grid map; acquiring distribution information of obstacles through the live-action three-dimensional model, judging whether the corresponding grid units are occupied by the obstacles or not, and generating an obstacle map according to a judgment result; dividing and storing space nodes in the grid map based on the obstacle map; performing initial path search by calculating a cost function of an initial node; and taking the obtained initial path as the basis of initial pheromone distribution, and obtaining an optimized path by adopting an ant colony optimization algorithm. According to the method, the influence of building shielding on image quality is avoided, time and energy waste of the unmanned aerial vehicle on an unnecessary path is reduced, the target image is obtained in the shortest time and the optimal path, and a flight task is completed.

The invention provides a path planning method, electronic equipment and a computer program product. The intelligent agent realizes path planning based on a pulse neural network of width, and the path planning method comprises the steps that a target area is divided based on environment information of the target area and an optimization target, a rasterized map composed of different sub-areas is determined, the optimization target comprises calculation efficiency and environment adaptability of the intelligent agent, and the rasterized map is determined based on the calculation efficiency and the environment adaptability of the intelligent agent. Each sub-region corresponds to one pulse neural network sub-node, and the rasterized map comprises a starting point position, an ending point position and obstacle information; based on the environment feature vectors of the sub-regions of the rasterized map and the position information of the intelligent agent, the current state of the intelligent agent is determined, and the current state is determined based on the Markov decision process; and based on the current state of the intelligent agent and a path planning algorithm, performing path planning on the rasterized map, and determining a planned path.

The invention discloses a forest pest and disease development trend prediction method and system combined with remote sensing monitoring, and relates to the related field of forestry monitoring data processing, and the method comprises the steps: carrying out the remote sensing monitoring of a target forest, building a remote sensing image data set, calling ground monitoring data, and carrying out the rasterization based on a time step after the interpolation; performing multi-feature joint extraction on the remote sensing image data set, and superposing extraction results to a unified grid to form a multi-layer feature map; constructing a joint feature matrix by using the rasterized data and the multi-layer feature map, and performing pest distribution identification through a classification model; performing pest development trend prediction by taking the pest distribution diagram as a basic feature and taking the combined feature matrix as an additional feature; and according to a development trend prediction result, carrying out pest abnormity early warning. The technical problem of insufficient prediction accuracy caused by insufficient monitoring efficiency and accuracy in existing forest pest development trend prediction is solved, and the technical effects of improving the monitoring efficiency and accuracy and improving the pest prediction accuracy are achieved.

The invention relates to a path planning method and system for navigation guidance of a low-altitude aircraft, and the method comprises the steps: calculating a probability density value of pre-processed prior flight data based on kernel density estimation, mapping the normalized probability density value to a grid map of a rasterized environment model, generating a flight path safety experience distribution map, and carrying out the navigation guidance of the low-altitude aircraft. And expanding the boundary of the obstacle area based on the safety distance constraint matrix, generating a restricted area matrix, generating an initial path according to the bidirectional A *-KDE algorithm cost function, the flight line safety experience distribution map and the restricted area matrix, removing redundant nodes from the initial path, and carrying out smoothing processing to obtain a flight route. The flight route is obtained by using the prior flight data, the deviation between the flight route and the actual navigation path can be reduced, the obstacle area boundary is expanded based on the safety distance constraint matrix, the obstacle safety distance is fully considered, and the risk that the low-altitude aircraft collides with the obstacle can be reduced.

The invention relates to the technical field of remote sensing image processing, and particularly discloses a photovoltaic power station remote sensing identification method and system based on time constraint, and the method comprises the steps: reading image data, and obtaining projection parameters and geographical transformation parameters; processing the image data by a minimum-maximum normalization method; selecting a reference year, acquiring reference year photovoltaic power station vector data processed by a normalization method, converting image data and vector data processed by the reference year normalization method into the same coordinate system according to projection parameters and / or geographical transformation parameters, rasterizing the vector data, and generating a binary mask consistent with the resolution of the image data; and by taking the binary mask as a label, designing a deep learning network structure to train image data, and outputting a predicted binary mask. According to the method, the uncertainty of single-phase image classification is remarkably reduced, and the stability and reliability of a multi-phase classification result are ensured.

The invention provides a self-adaptive texture atlas compression system and method based on a Unity engine, and relates to the field of graphic processing. The method comprises the steps that a texture resource import module converts PNG into a DYN file and generates configuration data, a texture resource setting module adopts a double-buffer configuration management mechanism, Hash value differences are detected through an improved CityHash algorithm, fields are changed asynchronously and synchronously, an image set construction module executes edge detection, polygon generation and rasterization processing, and a texture resource setting module performs texture resource setting. Pixel data is output by adopting an active side table algorithm, the arrangement unit realizes pixel data mutual exclusion arrangement based on multi-thread parallel calculation and outputs a UV coordinate mapping table, and an image set setting module dynamically modifies configuration parameters and triggers pipeline updating; geometric accuracy is ensured through verification of the number of vertexes and the texture area, expansion offset and step length parameters are dynamically adjusted to adapt to different resolutions, and an integrated breakpoint resume mechanism guarantees integrity of transaction logs.

The invention discloses an underground expressway reinforcement learning intelligent line selection method based on a GIS and an InSAR, and the method comprises the steps: providing urban environment multi-source data based on the GIS and the InSAR, building a rasterized observation space, building an initial linear continuous line space, and jointly building a line selection space of a reinforcement learning model; the strategy network of the reinforcement learning model receives parameter information of the line selection space and outputs linear parameters; the linetype builder corrects linetype parameters and updates the line selection space; the multi-index evaluation model outputs reward and punishment values according to the parameter information of the new line selection space, a value network of the reinforcement learning model receives the reward and punishment values, calculates a gradient for updating, and transmits a gradient updating strategy model; circulating until a round of line selection is completed, and calculating an accumulated reward value; and when the accumulated average reward value does not reach the optimization target, circulating until the optimization target is reached, and finding the optimal design. The method has the advantage of being beneficial to improving the efficiency and accuracy of underground road design.

The invention discloses an image alignment method and device, equipment and a storage medium, and relates to the field of image processing. According to the scheme, a to-be-detected circuit board scanning graph is obtained, and any unit area in the to-be-detected circuit board is recognized and extracted; based on a preset segmentation rule, segmenting the unit area into a plurality of subareas, and generating a window image set; and extracting window images in the window image set, mapping the window images to a mother layout corresponding to the Pcs particle image one by one according to image contours, and carrying out image alignment according to the image contours so as to carry out defect detection according to the mother layout. The method comprises the following steps: carrying out rasterization segmentation on a circuit board area image, and mapping the image to a target position in a mother layout in a manner that each window image is matched and aligned one by one; and each window image can be compared with the contour line of the mother layout as much as possible, the problems of element missing, dislocation, addition and damage are detected, and the subsequent defect detection precision is improved.

The invention relates to a vehicle driving control method, a vehicle and a computing platform, and the method comprises the steps: carrying out the three-dimensional point cloud data collection and all-round view image data collection of an environment where the vehicle is located, so as to obtain a three-dimensional point cloud frame and an aerial view image feature; rasterizing point clouds in the three-dimensional point cloud frame to obtain grid point clouds corresponding to a plurality of grids; determining a target deep learning model according to the distance between the grid and the vehicle; performing feature extraction on the grid point clouds in the grids by adopting a target deep learning model to obtain point cloud features of the grid point clouds; projecting the aerial view image features into a plurality of grids; fusing the aerial view image features and the point cloud features in the same grid to obtain fused features; identifying the fusion feature to obtain a target detection result of the three-dimensional point cloud frame and a classification detection result of the plurality of grids; and controlling the vehicle to run according to the target detection result of the three-dimensional point cloud frame and the classification detection result of the plurality of grids. And the accuracy of target detection is improved.

The invention discloses a vehicle target detection method based on binocular camera and laser radar fusion, and the method is characterized in that the method comprises the following steps: obtaining a left eye image and a right eye image of a vehicle target at a preset moment of the binocular camera, inputting the left eye image and the right eye image into a stereo matching network, and outputting a disparity map; inputting the false point cloud image into an image semantic segmentation network, segmenting a vehicle through the image semantic segmentation network, then generating a false point cloud only containing vehicle information, setting epsilon reliability and reducing the influence of noise by introducing an adaptive fusion factor delta and combining parallax prediction, and then fusing the true and false point clouds in a data set; a vehicle target is detected by rasterizing a vehicle target detection network, and the position, category and orientation of a three-dimensional target frame of a vehicle are obtained. According to the method, the precision of depth estimation, the effectiveness of the false point cloud and the precision of vehicle target detection are improved.

The invention provides a cross-platform collaborative loading method for CAD drawings, and relates to the technical field of data processing.The cross-platform collaborative loading method comprises the steps that rasterization layering processing is conducted on an original CAD file on a central platform, and a multi-stage grid tile set is obtained; storing to a temporary storage unit by adopting a pyramid hierarchical index; acquiring an updated view state by monitoring a user interaction event in real time, and comparing the updated view state with the original CAD file to obtain view parameters; extracting a target grid tile set; converting into a neutral exchange format set; according to the format support features, scheduling the container format conversion service, and compatibly distributing the container format conversion service to M edge platforms; and performing synchronous rendering replacement of the cross-platform tile file according to the M role permissions. The technical problems that in the prior art, due to the fact that effective CAD file format compatibility processing is lacked, tedious data conversion needs to be carried out among all platforms, and then the accuracy and reliability of data are reduced are solved.

The invention relates to the field of PCB detection, in particular to an AOI detection data processing system and method for a PCB, and the system comprises a visual processing module, a contour recognition module, a PCB alignment module, a region aggregation module and a defect detection module, the visual processing module is used for rasterizing each image pixel point, the contour recognition module is used for detecting a communication contour of the PCB, and the PCB alignment module is used for aligning the PCB alignment module. The PCB alignment module is used for aligning a PCB image with a standard image, the area aggregation module is used for identifying a defect view field and judging a qualified state, and the defect detection module is used for establishing a reinspection path and outputting a block defect rate. Production defects can be found in time, the probability of missing detection and false detection is reduced, the production quality of the PCB is improved, stable operation of an AOI system in a production environment is guaranteed, and the detection efficiency and accuracy are improved.

The invention discloses a large-scale building group layout optimization method based on deep learning. The method comprises the following steps: S1, obtaining a three-dimensional rasterized matrix and a meteorological environment vector of a to-be-optimized building group; s2, on the basis of a pre-trained three-dimensional building layout-wind field mapping network, generating a global wind environment distribution map according to the three-dimensional rasterized matrix and the meteorological environment vector; s3, based on the global wind environment distribution diagram, calculating a building layout parameter gradient through a gradient back propagation optimizer, and generating a layout optimization scheme; and S4, based on the layout optimization scheme, outputting optimized building group space layout parameters. Intelligent optimization of large-scale building group layout is realized through a deep learning technology, and urban wind environment quality and planning and design efficiency are remarkably improved.

The invention discloses a concrete strength detection method based on image recognition. The method comprises the following steps: generating a corrected concrete image data set; outputting a shared feature map set; obtaining a semantic mask set by adopting a rasterized pyramid decoder and a channel attention mechanism; performing microstructure feature statistics based on the semantic mask set to obtain a microstructure index set; outputting a concrete strength predicted value set through a transfer learning and fine tuning regression mechanism; performing incremental updating on the strength prediction network by using a pseudo label-active learning loop mechanism to obtain an updated strength prediction network; and reasoning the corrected concrete image data set by using the updated strength prediction network to generate a concrete strength visualization result map set. According to the method, the interference of a non-intensity correlation noise region on a segmentation result is effectively suppressed, and the accuracy of a microstructure index is remarkably improved.

The invention is suitable for the technical field of electric power circuit relocation and transformation digitization, and provides a digital electric power circuit relocation and transformation design method which comprises the following steps: rasterizing a target relocation and transformation area, and performing cost assignment on each grid to obtain the cost expectation of each grid; according to the cost expectation of each grid, a low-cost area composed of low-cost grids in the target migration and modification area is determined; determining a starting grid and a target grid of a target path in the target relocation and relocation area based on the existing power line and the target relocation and relocation area; and searching a plurality of migration and modification paths between the starting grid and the target grid in the target migration and modification area by using an ant colony algorithm which introduces the low-cost area. The invention further provides a digital power line relocation design system. Through combination of rasterization processing, cost-benefit analysis, an intelligent algorithm and decision support, an efficient and economical solution can be provided for an electric power line relocation project.

The invention provides an electric field analysis method. Firstly, a target model is generated through simulation modeling, continuous electric field distribution data is obtained, a first data set and a second data set are established, rasterized image feature extraction is performed on the first data set and the second data set through a graph neural network, corresponding first graph features and second graph features are obtained, and a cascade grid processing network is used to process the electric field distribution data. Firstly, a target power line is predicted according to the first graph feature, then the target power line is converted into a target electric field mode according to the second graph feature, and finally an electric field prediction result is generated and a preset analysis process is executed. According to the method, through the combination of the graph neural network and the cascade grid processing network, the continuity characteristic of electric field distribution is effectively reserved, and meanwhile, the noise problem caused by discretization is avoided. And through rasterized feature data processing, the data demand and the calculation complexity are greatly reduced, and the efficiency and the accuracy of modeling analysis are improved.

The invention relates to the technical field of ballistic calculation, and discloses a rapid ballistic calculation method and system, computer equipment and a storage medium, and the method comprises the steps: generating ballistic calculation parameters based on rasterization, carrying out calculation, obtaining multiple pieces of original ballistic data, carrying out the feature extraction and segmentation processing of the original ballistic data, and obtaining a sample data set; the sample data set is adopted to train a trajectory resolving model, a trained trajectory resolving model is obtained, and the trajectory resolving model comprises a front section trajectory sub-model, a middle section trajectory sub-model, a rear section trajectory sub-model and a trajectory feature sub-model; and inputting to-be-calculated parameters into the trained trajectory calculation model for calculation to obtain trajectory scatter data. According to the method, the deep learning technology and the actual ballistic segmentation characteristics are combined, the rapid ballistic prediction method based on the segmentation training strategy is provided, the rapid ballistic calculation at the millisecond level is achieved on the premise that the prediction precision is guaranteed, and the requirements for missile combat mission planning and decision real-time performance are met.

The invention relates to a model training method, a map data fusion method and related equipment, and relates to the technical field of automatic driving. The model training method comprises the following steps: acquiring at least two local vector maps and global map data; extracting the at least two local vector maps to obtain category data and coordinate data of a plurality of map elements; performing feature extraction on the category data and the coordinate data by using the initial map data fusion model to obtain vector map features; performing rasterization processing based on the coordinate data to obtain raster map features; performing feature fusion and feature extraction on the vector map features and the grid map features to obtain extracted map features; decoding and predicting the extracted map features to obtain global map prediction data; and training the initial map data fusion model based on the global map prediction data and the global map data to obtain a map data fusion model. According to the invention, an accurate fusion result can be obtained, and the fusion efficiency is high.

According to the intelligent navigation method and device for the inspection robot, a map fusion system is innovatively designed, and the environment is effectively expressed through multi-source data and rasterization processing. And an environment perception mechanism is constructed, and a reliable obstacle recognition model is established in combination with semantic analysis and trajectory prediction. Dynamic planning is introduced, and the accuracy of obstacle avoidance is ensured through interference analysis and state optimization. According to the method, the defects of the traditional technology in the aspects of map fusion, environment perception, dynamic planning and the like are effectively overcome, and technical guarantee is provided for the inspection robot.

The present application provides a method and device for fusing traffic data, the method comprising: obtaining multiple traffic data sets collected from a target road section, and rasterizing two of the traffic data sets to be fused respectively, to obtain a spatiotemporal traffic map corresponding to each traffic data set to be fused; based on the data source corresponding to each traffic data set to be fused, fusing the spatiotemporal traffic map to obtain a target spatiotemporal traffic map of the target road section, and determining at least one first target congestion coordinate set in the target spatiotemporal traffic map through morphological operations and connected domain analysis; or fusing at least one congestion coordinate set corresponding to each spatiotemporal traffic map to obtain at least one second target congestion coordinate set corresponding to the spatiotemporal traffic map of the target road section. Through the above method, the accuracy and reliability of traffic data fusion are improved, thereby improving the efficiency of traffic management.

The invention discloses an isomorphic unmanned aerial vehicle global situation information sharing method and system, and the method comprises the steps: constructing a rasterized dynamic situation map, and initializing the rasterized dynamic situation map; issuing and planning to-be-searched area information and target information through the situation map, and synchronously updating the situation information to other unmanned aerial vehicles in the cluster; the unmanned aerial vehicle performs target characteristic matching and updates a situation area in the situation map; planning a cluster unmanned aerial vehicle flight path according to the updated situation map, and allocating the unmanned aerial vehicles to execute a tracking task and a region search task; and in the task execution process of the unmanned aerial vehicle, distributed situation map targets are added, updated and deleted. According to the invention, global real-time co-construction and sharing of situation information are realized, perception delay caused by forwarding of the situation information in a centralized architecture is avoided, the ability of the unmanned aerial vehicle cluster to perceive the scene situation more rapidly is realized, and the unmanned aerial vehicle cluster is assisted to realize autonomous rapid formation and autonomous task decision in a task area.

The invention provides a document image-text integrated intelligent understanding and processing method and system, and relates to the technical field of artificial intelligence. Comprising the following steps: separating image-text elements and establishing context association; calling a graph vectorization engine, and executing intelligent vector conversion processing on the rasterized illustration; performing hierarchical classification, attribute endowing and structured reconstruction on the vector primitives in combination with text semantic analysis to generate vector entity objects with complete attributes; the text elements and the vector entity objects are jointly input into a pre-trained multi-mode large language model, unified knowledge expression containing vectors, topology, attributes and document contexts is generated, and applications such as question answering, editing, abstracting and reporting of image-text integration are supported. According to the method, the problems of graph semantic loss, difficulty in fine analysis and the like caused by vector-to-grid illustration in papers, reports and other types of documents are solved, the overall understanding depth and interpretability of AI for complex image-text documents are remarkably improved, and high-performance semantic understanding capacity is provided for document-based training and intelligent processing.

The invention relates to the technical field of numerical control machining measurement and correction, and discloses a machining surface correction method for a numerical control milling center. The method comprises the steps of establishing an original point cloud of a processing surface and extracting a feature structure; analyzing each feature structure to form a structured description list; inputting the list into a geometric growth model, and generating a virtual feature structure set according to rule iterative evolution; extracting and rasterizing the contour and the height field, carrying out registration fusion with a reference plane equation, and carrying out variation calculation to obtain a smooth middle geometric surface; performing adaptive resampling on the intermediate surface to generate an intermediate point cloud; and finally, according to the spatial frequency distribution of a topological residual field formed by the difference between the original point cloud and the intermediate point cloud, adaptively determining a weight, and carrying out weighted mixing on the original point cloud and the intermediate point cloud to generate an optimized surface point cloud. According to the method, intelligent evolution of the machined surface features and frequency domain self-adaptive data fusion are achieved, and the correction machining precision and the surface forming quality are improved.

The invention provides a road boundary detection method and a false target identification method and system, and the method comprises the steps: detecting the operation information of a vehicle in real time, and updating the historically accumulated point cloud information based on the motion information of the vehicle; performing screening processing on the point cloud information based on a preset screening rule to obtain final point cloud information; performing rasterized indexing on the final point cloud information based on a preset rasterized indexing rule to obtain a preliminary road boundary point; and based on the self-vehicle motion information and the final point cloud information, carrying out smoothing processing on the initial road boundary point to obtain a final road boundary. According to the method provided by the invention, the technical problem that the road edge identification technology cannot accurately identify the road boundary in special-shaped road boundaries or irregular road scenes, such as forks, roundabout, small roads in villages and the like, is effectively solved.

The invention discloses a thrown object detection method and system based on a traditional algorithm and deep learning. The method comprises the following steps: sequentially carrying out background filtering, category labeling, data enhancement, rasterizing operation and rasterizing partitioning, ground filtering and point cloud compensation on an original point cloud set to obtain a post-processing point cloud set; performing model training by taking the post-processing point cloud set as a training sample, and introducing a terrain clearance factor to obtain a target point cloud segmentation network model; performing background filtering, ground filtering and point cloud compensation on the to-be-detected point cloud in sequence to obtain to-be-detected post-processing point cloud; inputting the post-processing point cloud to be detected into the target point cloud segmentation network model for prediction to obtain the category of each point; and extracting the points of which the category is the throwing object, and calculating an external frame of the points to obtain the center position and the length, width and height size information of the throwing object. The method provided by the invention can accurately detect the small target thrown object, and improves the detection precision while ensuring the detection efficiency of the thrown object.