324results about How to "Reduce calculation" patented technology

The invention discloses a lightweight dynamic security authentication method and system based on PUF (Physical Unclonable Function), and the method comprises the steps: S1, a server side collects to-be-registered information, generates a random challenge and auxiliary information, receives a response corresponding to the challenge generated by an IOT (Internet of Things) device side through a PUF structure, and sends the response to the server side; storing the pseudonym and the private key of the registered user and the CRP pair of the user equipment; s2, when the server side receives the authentication request, the server side initializes the authentication request, generates a challenge of the next round of authentication by using the challenge of the current authentication in the initialization process, and sends the challenge of the next round of authentication to the Internet of Things equipment side; the server side verifies the identity of the device after receiving the data packet sent by the Internet of Things device side; and S3, updating a next round of pseudonym at the server side, generating a next round of challenge and CRP pair by the server side, and sending the challenge and CRP pair to the Internet of Things equipment side. The method has the advantages of being simple in implementation method, small in resource overhead, high in flexibility, capable of dealing with various security threats and the like.

The invention belongs to the technical field of model training, and particularly relates to a large model increment training method and system based on dynamic sparsification. The method comprises the following steps: S1, evaluating the parameter importance of a large model by adopting a multi-dimensional evaluation method, and reflecting the contribution of parameters to the performance of the large model; s2, performing dynamic sparsification control on the large model structure by adopting a self-adaptive structure adjustment method; s3, performing incremental training management on the large model by constructing a capacity-adjustable empirical playback buffer pool and proposing a multi-objective loss function; and S4, carrying out model compression optimization on the large model by adopting a mixed precision quantification strategy. The method has the characteristics that the network structure can be dynamically adjusted by evaluating the importance of model parameters in real time, so that the efficient training is realized while the model performance is ensured.

The invention belongs to the field of computer graphics, and discloses a fast parallel multi-scale smooth symbol distance curved surface reconstruction method and system, and the method comprises the steps: carrying out the local fitting of point cloud data through employing a moving least square method, and calculating the Gaussian curvature of a point cloud; constructing a curvature adaptive octree based on Gaussian curvature, wherein the octree is used for dynamically adjusting local resolution and capturing geometric details to improve curved surface reconstruction precision; global fitting is carried out through an implicit function, and a zero contour surface is obtained through an octree contour surface extraction algorithm; and multiple threads are introduced in the process of iteratively solving the linear system to carry out parallel computing. The adaptive octree based on Gaussian curvature can effectively capture complex geometric details, and the algorithm realizes high-precision curved surface reconstruction under different scales. In addition, the local and overall calculation efficiency is remarkably improved through parallel calculation, and the high efficiency of the algorithm in large-scale data processing is ensured. The method provided by the invention provides an efficient and accurate solution for complex curved surface reconstruction.

The invention discloses an electric power system state estimation method based on graph contrast learning. The method comprises the following steps: constructing graph sequence data of electric power system state estimation; constructing an electric power system state estimation model based on graph contrast learning, and putting the graph sequence data into the electric power system state estimation model based on graph contrast learning; inputting the node feature sequence and the masked edge features into a hybrid encoder of a dynamic edge convolutional neural network combining a self-attention mechanism and a residual block and a graph convolutional neural network with a dynamic propagation strategy of a gating mechanism to comprehensively process the features of nodes and edges; comparing an updating mechanism and a sampling symmetry training strategy of model parameters of the learning framework through a graph; and calculating a node voltage phase angle by using the branch phase angle difference, and outputting a state estimation result. According to the method, the internal structure and mode of the power system data can be effectively learned by combining graph contrast learning and a deep neural network technology, and the capability of identifying and filtering abnormal data is improved.

The invention provides a multi-degraded image restoration method based on self-adaptive prompt. The method comprises a two-stage hidden space diffusion model, and various degraded images can be flexibly processed. Specifically, the model generates a degradation cue by learning the difference between a degraded image and a clean image, and encodes the cue information in a hidden space. In the first stage, a degradation prompt is obtained through a degradation adaptive prompt learning mechanism as a key feature for guiding the diffusion model in the second stage to carry out image restoration. Finally, prompt information is integrated into a Fourier transform restoration network, and accurate restoration of different types of degraded images can be realized. The method not only improves the accuracy of image restoration, but also significantly reduces the calculation complexity, and is suitable for image restoration tasks in various real world scenes, including denoising, deblurring, defogging, rain removing, snow removing, raindrop removing and low illumination enhancement.

The invention discloses an unmanned aerial vehicle target detection method based on feature fusion and detection head collaborative optimization, and the method comprises the steps: obtaining a multi-scale feature map through employing a backbone network formed by stacking a convolution module CBS and a residual module C2f, sequentially transmitting the feature maps of various scales to a multi-level feature fusion module (ML-FFM), carrying out the feature fusion from a deep layer to a shallow layer, and carrying out the detection of a target target. Semantic information of shallow layer features is enhanced; inputting the fused feature map into a refined feature enhancement module (RFEM), eliminating position conflicts and redundant information based on a space-time attention mechanism, and optimizing details and positioning capability of deep features; and the enhanced feature map is sent to a prediction head based on Transform for detection, and target classification and positioning of the aerial image of the unmanned aerial vehicle are completed. According to the method, the problems that the unmanned aerial vehicle aerial image is large in scale change, and small targets are high in proportion and dense in distribution are solved, so that the precision of target detection in the unmanned aerial vehicle aerial image is improved. The modules proposed in the method can significantly reduce the calculation amount, and are suitable for the real-time detection scene of the unmanned aerial vehicle.

The invention relates to the technical field of image recognition, in particular to a circuit board defect detection method and device, equipment and a storage medium, and the method comprises the steps: obtaining a to-be-detected image, and carrying out the preprocessing of the to-be-detected image, and obtaining a preprocessed image; inputting the preprocessed image into a pre-constructed defect detection model, performing feature extraction processing on the preprocessed image based on a pre-confirmed convolution weight, performing fusion processing on a plurality of feature maps of different scales based on a pre-confirmed fusion weight, and performing defect detection on the fused feature map by adopting a detection head to obtain a defect detection result; the defect detection method disclosed by the invention aims to solve the problems of high calculation complexity, poor real-time performance and insensitivity to small target detection in the existing PCB defect detection method. The convolution weight is adjusted by introducing a dynamic convolution selection mechanism, and it is ensured that the model has high precision when processing small defects; and the fusion weight is adjusted by introducing a weight adaptive mechanism, so that the calculation of redundant information can be reduced, and the detection efficiency is greatly improved.

The invention relates to the technical field of robot control instruction analysis, in particular to a robot control instruction analysis system based on multi-algorithm fusion, which comprises a data collection module, an environment map analysis module, a global path analysis module and a moving trajectory analysis module, wherein the data collection module collects state data of a robot, surrounding environment data of the robot and a target position; the environment map analysis module determines an environment occupation grid map according to the environment data; the global path analysis module determines a global path according to an environment occupation grid map, a target position, self state data and robot motion constraint parameters; the running track analysis module determines a smooth track according to the global path and the state data of the running track analysis module and detects an environment occupation grid map, if a dynamic obstacle avoidance object exists in the environment occupation grid map, the environment occupation grid map is updated, the smooth track is modified according to the environment occupation grid map updated in real time, and the dynamic obstacle avoidance object is obtained. And the final robot moving trajectory is obtained.

The invention discloses an optical perspective calibration method and system, electronic equipment and a storage medium, and relates to the field of augmented reality. In the method, the position and posture of a virtual calibration object on a display screen are adjusted according to received first user voice, so that the virtual calibration object coincides with an actual calibration object in a physical space; obtaining an image of an actual calibration object according to the received second user voice, obtaining a first two-dimensional pixel coordinate of the first target corner point from the image, and determining a three-dimensional coordinate of the first target corner point according to the first two-dimensional pixel coordinate; acquiring a second two-dimensional pixel coordinate of a second target angular point of the virtual calibration object on the display screen, and calculating a projection matrix according to the three-dimensional coordinate and the second two-dimensional pixel coordinate; the projection matrix is used to transform the points in the virtual space onto the screen of the augmented reality headset. By implementing the technical scheme provided by the invention, personalized optical perspective calibration is realized in a low-cost manner.

The invention belongs to the technical field of unmanned aerial vehicles, and particularly provides a transformer substation unmanned aerial vehicle autonomous inspection route generation method based on an optimization A star algorithm. According to the method, a heuristic function is improved, the Euclidean distance between a target point and a current node is comprehensively considered, obstacle avoidance processing is carried out, and a safer and more efficient inspection route is generated. And by calculating the total cost of each node, the node with the minimum total cost is selected for expansion until a target node is found. According to the improved A star algorithm, the unmanned aerial vehicle can be effectively prevented from flying into an obstacle dense area, the obstacle keeps a safe distance, and the flying safety is remarkably improved. The distance and the obstacle density are comprehensively considered, the generated inspection route is smoother and more efficient, and unnecessary flight distance and energy consumption are reduced.

The invention relates to a deep learning algorithm, and provides an end-to-end sparse trajectory recovery method based on a road network, which is realized based on a framework of a generator and a discriminator, and is characterized in that the generator recovers an input sparse trajectory into a complete trajectory with a high sampling rate; the discriminator evaluates the authenticity of the input track and is used for guiding the generator to update parameters; the fusion of trajectory recovery and map matching is realized in an end-to-end manner through a deep learning technology, and parameter updating of a generator is optimized through a GAN (Generative Adversarial Network) framework, so that the balance between the completeness of the recovered trajectory and the accuracy of single-step interpolation is realized. According to the invention, the precision and speed of track recovery can be obviously improved.

The invention discloses an unmanned aerial vehicle assisted Internet of Vehicles cross-domain authentication method based on a block chain and federated learning, and the method comprises the steps: initializing an Internet of Vehicles system, and generating system initialization common parameters; the vehicle and the unmanned aerial vehicle complete identity registration with the control center, and the vehicle performs login verification; and respectively carrying out initial authentication and cross-domain authentication on the vehicles which are successfully logged in and verified, guaranteeing the communication security through key negotiation, and completing the cross-domain authentication of the unmanned aerial vehicle assisted Internet of Vehicles. According to the method, the reliability and the security of data in the cross-domain authentication process are realized, the privacy of the data is improved, relatively low calculation and transmission overhead in the cross-domain authentication process is realized, and the authentication efficiency is improved.

The invention relates to the technical field of scene recognition, in particular to a real-time semantic segmentation method and system for a multi-shape pyramid in a traffic scene, and the method comprises the steps: initializing an original to-be-segmented image, and obtaining an initial segmentation feature map; channel splicing is carried out on the initial segmentation feature map and an output feature map obtained after first downsampling is carried out on the original to-be-segmented image, downsampling is carried out after a fused feature map is obtained, detail information is extracted, and a detail information feature map is obtained; down-sampling the detail information feature map and extracting semantic information; the method comprises the following steps: extracting transverse and longitudinal receptive field information of a semantic information feature map, performing expansion convolution to obtain feature maps of different receptive fields, obtaining a multi-scale and multi-shape context information feature map, performing up-sampling, performing point-by-point convolution dimension reduction on a detail information feature map, and performing channel splicing on the two features to obtain a fusion feature map containing semantic and detail information. And adjusting the channel number into a prediction category number, and normalizing to obtain a prediction segmentation result. The method has good segmentation precision and reasoning speed.

The invention discloses a lightweight operating equipment anomaly detection method and system, and the method comprises the steps: collecting vibration signal data of operating equipment, carrying out the preprocessing of the vibration signal data, and enabling the preprocessed data to serve as a data set; inputting the data set into the constructed isolated forest model for anomaly detection, labeling the data set according to an anomaly detection result, and taking the labeled data set as a neural network training data set; constructing an operation equipment anomaly detection model, and training the operation equipment anomaly detection model by using the neural network training data set; converting the trained model into a lightweight operation equipment anomaly detection model; importing the lightweight model into a micro-control unit of the power consumption embedded vibration sensor; and collecting vibration signal data of to-be-detected operation equipment in real time, performing data preprocessing on the vibration signal data, and inputting the vibration signal data subjected to data preprocessing into the lightweight operation equipment anomaly detection model in real time to perform operation equipment anomaly detection. The method has the advantages of low cost, high prediction accuracy, easiness in implementation and the like.

The invention discloses a transformer substation scene three-dimensional semantic segmentation method based on combination of convolution and attention, and the method comprises the following steps: S1, carrying out the region and category division of three-dimensional point cloud data in a transformer substation scene, setting semantic tags for different categories, dividing a training set and a test set according to the region, and forming a three-dimensional semantic segmentation data set; s2, constructing a three-dimensional semantic segmentation model based on a convolution and attention mechanism combination module, wherein the three-dimensional semantic segmentation model is used for predicting a semantic tag of point cloud data in a substation data set; s3, determining a loss function and a parameter optimizer, and training a three-dimensional semantic segmentation model in combination with the three-dimensional semantic segmentation data set; and S4, performing semantic segmentation on the point cloud data in the substation scene based on the trained three-dimensional semantic segmentation model. Compared with an existing three-dimensional semantic segmentation method, the accuracy of three-dimensional semantic segmentation can be effectively improved, and accurate semantic segmentation of the substation scene point cloud data is achieved.

The invention relates to a high-resolution remote sensing image building function type classification method based on a multi-hop graph neural network, and belongs to the field of remote sensing image classification. The method comprises the following steps: firstly, constructing a CNN-Transform double-encoder semantic segmentation model to extract building area information, then, constructing graph structure data based on the building area information, and then, constructing a multi-hop graph neural network to identify a building function type; according to the CNN-Transform double-encoder semantic segmentation model, local space detail features are kept while global context features are extracted; according to the multi-hop graph neural network, the multi-hop neighborhood features of the nodes are calculated in advance, and the multi-hop graph neural network can be used for large-scale graph structure data only by calculating once in a CPU or a distributed system. The network supports small-batch training and is easily expanded to a large data set, the complexity and calculation cost of node interaction in the training and prediction process are eliminated, the accuracy and expansibility of building function type classification are enhanced, and technical support is provided for efficiently obtaining building function types in remote sensing images.

The embodiment of the invention provides a training method and device of a text abstract generation model, equipment and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: acquiring a sub-word index sequence of a text sample; processing the sub-word index sequence to obtain a coding tensor sequence; according to the coding tensor sequence, carrying out internal loop training and external loop training on a preset TTT-Learar-Transformer model, and obtaining a target loss function; according to the target loss function, weight updating is carried out on the TTT-Learar-Transform model in a dual form until the TTT-Learar-Transform model converges, a text abstract generation model is obtained, and the text abstract generation model is used for generating an abstract of a text. According to the embodiment of the invention, the performance of the text abstract generation model can be improved, and the text abstract generation quality and efficiency are improved.

The invention belongs to the technical field of image processing, and discloses a loess landslide detection method and system optimized by a multi-population grey wolf optimization algorithm based on grade differentiation learning. Performing data enhancement on the landslide data set by using a data enhancement generator; designing a grey wolf optimization algorithm model based on multi-level differential learning through the segmentation network model, and optimizing hyper-parameters in the segmentation network model by using an improved algorithm to realize adaptive adjustment and cooperative calculation of three hyper-parameters, namely a learning rate, an iteration number and a batch processing number, in the segmentation network; and finally carrying out loess landslide detection. According to the method, algorithm improvement is carried out, so that the balance capability of global exploration and local development is enhanced, and later diversity loss and local optimum are avoided. By combining the excellent semantic segmentation capability of VIG and the improved grey wolf optimization algorithm, the method can effectively overcome the feature extraction difficulty caused by factors such as geological structure and vegetation coverage, and significantly improves the accuracy of loess landslide recognition.

The invention provides a fuzzy privacy set intersection method and system, a medium and electronic equipment, and the method comprises the steps: carrying out the encryption processing of an initial data set X through a homomorphic encryption algorithm, obtaining first data, and obtaining a processed data structure D through a defined space Hash algorithm; the receiver performs homomorphic addition or scalar multiplication operation processing on the first data by using own data Y, and performs confusion processing by using a random number R to obtain second data; after the sender carries out Augment algorithm processing on the random number, a Hash algorithm is carried out to obtain a corresponding Hash value, and the Hash value is recorded as third data to be sent to the sender; and the sender decrypts the second data, processes the second data through an Augment algorithm, executes a PSI protocol with the receiver to obtain fourth data, and compares the third data with the fourth data to obtain intersection data. According to the method, an addition homomorphic technology is adopted for a novel application scene of intersection of the fuzzy privacy set, and compared with a full homomorphic technology in the prior art, the time consumption is less, and the communication is small.

The invention discloses a fine tuning method and system for precise parameters of a pre-training language model. The method comprises the following steps: constructing a robust adapter; a plurality of constructed robust adapters are connected in series to form a robust adapter chain, and the maximum chain node is determined; preparing a weight of a pre-training language model and a data set of a corresponding task label; firstly, each robust adapter in the robust adapter chain is subjected to preliminary training, optimization and approximate fine tuning of parameters of the robust adapters; after the parameters of the robust adapter are approximately fine-tuned, the parameters are combined with the weight of the preorder frozen pre-training language model to complete calculation of a current robust adapter chain node, residual information used for next iteration is generated, and the parameters of the current robust adapter are updated at the same time; and carrying out iterative training, updating residual information until the number of iterations is equal to the maximum node of the robust adapter chain, and obtaining the optimal weight of the pre-training language model and the optimal parameters of the robust adapters, so that more stable convergence and simpler hyper-parameter adjustment can be provided.

The invention belongs to the technical field of flood scene simulation, and particularly relates to a river flood routing grid model dynamic display method based on a Cesium three-dimensional scene, a readable storage medium and a program product. The method comprises the following steps: firstly, preprocessing Cesium three-dimensional scene geographic data for Cesium three-dimensional scene construction work; secondly, utilizing a two-dimensional hydrodynamic model of HEC-RAS software to construct an unstructured river flood routing grid model; then, converting a calculation result of the river flood routing grid model into a data format which can be identified by a Cesium three-dimensional scene; then, processing the preliminarily established unstructured river flood routing grid model again, and converting the preliminarily established unstructured river flood routing grid model into a structured river flood routing grid model; and finally, displaying the constructed river flood routing grid model in a Cesium three-dimensional scene. According to the method, the dynamic visualization effect and the rendering efficiency of the river flood routing are effectively improved.

The invention discloses a multi-objective optimization energy scheduling method for an active power distribution network, and the method comprises the following steps: building an index evaluation system including scheduling cost, credit and power supply guarantee, and carrying out the evaluation of the schedulable potential of a new energy vehicle, a flexible load, an energy storage device and distributed energy during different scheduling periods. Determining the dispatching priority of the power distribution network by adopting a method of combining an analytic hierarchy process and an entropy weight method; a multi-target scheduling optimization model and constraint conditions are established in the aspects of scheduling economy, renewable energy consumption capacity and user satisfaction, and the convergence optimization capacity and speed of a multi-target optimization algorithm are improved based on an improved particle swarm optimization algorithm. According to the method, the scheduling priority can be searched in the global active power distribution network to ensure that the optimal vector solution is obtained by the multi-objective functions of economy, absorption rate and satisfaction, so that the scheduling reasonability of the active power distribution network under the condition of encountering load demand response is improved, and the method is simple, efficient, easy to implement and good in practicability and economy.

The invention discloses a double-tower system process optimization method based on LU decomposition, and the method comprises the steps: taking the total value of a liquid phase product produced by a double-tower system formed by a depropanizing tower and a debutanizing tower as an optimization target, and constructing an optimization model by combining a constraint equation and a normalization equation; and performing iterative solution on the optimization model based on LU decomposition and a quadratic sequence programming method. According to the method, by judging the correlation of the constraint equation and optimizing the condition number, the influence of ill-conditioned problems can be effectively avoided, and the numerical stability in the solving process is remarkably improved. According to the method, the redundant constraint equation is removed, and the condition number is improved through LU decomposition basis changing, so that unnecessary calculation can be reduced, and the convergence speed of iteration is improved.

The invention discloses a truth value mining method for a multi-source unstructured text service in a meta-universe crowdsourcing environment, and provides a solution of three stages, namely a semantic preprocessing stage, a high-dimensional content vector representation of a crowdsourcing answer is constructed, and clustering is carried out; in the truth value optimization and feature generation stage, the optimized truth value mining model is used for evaluating the work quality of crowdsourcing workers, and features are dynamically generated. And a task clustering and class cluster mapping stage: clustering the tasks with generated features, accurately estimating the real class of each task, and finally determining a true value. According to the method, the accuracy and efficiency of true value mining are improved, and the applicability of the method in a complex unstructured data environment is expanded. By integrating the multi-source sensing data in the meta-universe crowdsourcing environment, the intelligent service meeting the user requirements can be accurately pushed, and the service quality and the user experience of the meta-universe are improved.

The invention discloses a rapid batch switching authentication method for predictability and aggregation promise proof, which realizes batch authentication of a plurality of user terminals by generating system parameters and secret keys through a trust mechanism and combining predictability promise and aggregation signature technologies. In the authentication process, the edge base station to be switched exchanges with the current communication edge base station through an encrypted message, verification information such as a pre-authentication certificate and a signature is generated through the user terminal, and identity authentication is completed step by step. The method comprises the following specific steps: a trust mechanism generates a private key, a public key, a pseudo identity and an expiration timestamp for each device; through multiple rounds of message exchange, the edge base station and the user terminal perform identity verification by using a shared key and a hash function; and finally carrying out efficient verification on a plurality of users through batch signature and aggregation commitment. According to the method, on the basis of ensuring the security, the speed of switching the authentication process is greatly improved, and the method is suitable for a large-scale network environment, especially in edge computing and mobile communication networks.

The invention discloses a virtual power plant optimal scheduling method and device, a medium and a product, and relates to the technical field of power plant electric energy resource scheduling and allocation, and the method comprises the steps: obtaining basic data of a virtual power plant; respectively carrying out clustering analysis on the temperature control load data and the electric vehicle data by adopting a density clustering algorithm and a K-means + + algorithm based on an HHACOA algorithm to obtain a clustering result; according to the basic data of the virtual power plant, an internal approximation constraint space integration method is adopted, and adjustable power domain data of the whole temperature control load cluster and adjustable power domain data of the whole electric vehicle cluster are determined; establishing a planning function optimization model according to the adjustable power domain data of the temperature control load cluster, the adjustable power domain data of the electric vehicle cluster and the clustering result; and solving the planning function optimization model to obtain an optimization strategy result of the virtual power plant. The method can improve the adjustment capability of the virtual power plant to the power grid, and guarantees the safety and stability of the power system.

The invention discloses a method for realizing capacitor voltage balance of a four-level energy storage converter under a low modulation ratio condition, and the method comprises the steps: converting a space vector diagram of the four-level energy storage converter under an alpha-beta coordinate system into a g-h coordinate system at a low modulation ratio, and obtaining three nearest vectors for synthesizing a reference voltage vector Vref and duty ratios of the three nearest vectors; therefore, the sector judgment process is simplified, and the calculation process of the duty ratio acted by the basic voltage vector is reduced. Then, according to the small triangle where the reference voltage vector Vref is located and the duty ratio of the action of the three vectors for synthesizing the reference voltage vector Vref, a five-segment type modulation synthesis mode and a seven-segment type modulation synthesis mode are adopted, and a switch sequence corresponding to each small sector is obtained; and finally, the switching sequence corresponding to the small sector is evaluated through the target function, and the switching sequence when the target function is the minimum value is selected as the optimal switching sequence by using comparison operation. And capacitor voltage balance of the four-level energy storage converter is realized. The technical problem that an existing additional auxiliary circuit balancing method is poor in practicability can be solved.

The invention belongs to the technical field of Internet security, and discloses a public Internet of Things privacy protection identity authentication method and system based on a block chain, which do not depend on a centralized server, adopts a storage mechanism based on a distributed hash table to realize data storage in a chain-on-chain and chain-off mode, realizes decentralized storage and improves the security of data. Identity sources of the sensor nodes are authenticated through an intelligent contract technology under the condition that real identities are not needed; the method has unforgeability under a random oracle model, and meets the security requirements of integrity, authentication, anonymity and traceability at the same time.

The invention discloses a federal recommendation privacy protection method based on ring signature and selective aggregation, which realizes anonymization privacy protection and selective aggregation in a federal learning process through an innovative double-server architecture and a hierarchical security mechanism. The method comprises the following steps: 1) deploying an aggregation server and a security server in a system initialization stage, and completing parameter initialization and key distribution; 2) the client executes local model training, and anonymization privacy protection uploading is realized by adopting a ring signature technology; 3) the security server performs selective aggregation of credibility verification, and completes grouping and parameter fusion based on user similarity; and 4) ensuring the completeness and source credibility of model updating through a hierarchical security parameter distribution mechanism. While the recommendation precision is ensured, the user identity leakage and privacy tracking are effectively prevented, the comprehensive security guarantee is provided for the federal recommendation system, and the method has the advantages of low calculation overhead and high communication efficiency.

The invention relates to the technical field of image processing, in particular to a marine metal corrosion degree detection method based on image processing, and the method comprises the steps: obtaining a plurality of grey-scale maps of a marine metal sample, taking the probability that a pixel point belongs to a background pixel point as a first probability, dividing a target pixel point and the background pixel point based on the first probability, taking the probability that the target pixel point belongs to the analysis pixel point as a second probability, determining an analysis area according to the second probability, and calculating the probability that the edge pixel point belongs to an inner edge pixel point as a third probability according to the fluctuation characteristics of the gray values of the edge pixel point of the analysis area and the adjacent pixel point in the gradient direction of the edge pixel point, inner edge pixel points are determined according to the third probability, the inner edge pixel points are traversed in the main direction, shadow areas are jointly determined according to the traversed pixel points and preset conditions, the corrosion degree is determined according to the area ratio of all the analysis areas to all the shadow areas, and the accuracy of marine metal corrosion degree detection is improved.