91 results about "Trajectory clustering" patented technology

The invention discloses a method for identifying the video action based on trajectory sequence analysis and rule induction, which solves the problems of large labor intensity. The method of the invention divides a complete trajectory in a scene into a plurality of trajectory section with basic meaning, and obtains a plurality of basic movement modes as atomic events through the trajectory clustering; meanwhile, a hidden Markov model is utilized for establishing a model to obtain the event rule contained in the trajectory sequence by inducting the algorithm based on the minimum description length and based on the event rule, an expanded grammar analyzer is used for identifying an interested event. The invention provides a complete video action identification frame and also a multi-layer rule induction strategy by taking the space-time attribute, which significantly improves the effectiveness of the rule learning and promotes the application of the pattern recognition in the identification of the video action. The method of the invention can be applied to the intelligent video surveillance and automatic analysis of movements of automobiles or pedestrians under the current monitored scene so as to lead a computer to assist people or substitute people to complete monitor tasks.

The invention relates to a target tracking method based on character feature invariant and graph theory clustering which overcomes the defects of the existing target tracking method, tackles the difficult problems of target scale changes, rotations, noises, diurnal variations, shadings, conglutinations, camera vibrations and the like in a scene and generates the stable target trajectory and accurate motion information thereof. The tracking method comprises the following steps: demarcating a camera; preprocessing an image; detecting a character; extracting invariant features, calculating invariant features at angular points; matching features, performing invariant feature matching between one angular point of the previous frame and all the angular points of the neighborhood of the local frame; forming angular point trajectories, connecting the frame-matched angular points to form the trajectory of the angular point; performing trajectory clustering based on graph theory, forming a plurality of temporary targets after clustering; combining or splitting targets, determining that the target and the temporary targets obtained by clustering combine or split, performing reasonableness test to update the current set target; performing reasonableness test to judge the reasonableness of the trajectory and scale of the target; and extracting Gaussian background and angular point background.

The invention provides a fast updating method of road network based on trajectory adaptive clustering, characterized in that the method comprises: whether the trajectory points match with the originalroad network is judged by the distance constraint condition and the direction constraint condition between the collected moving trajectory data and the acquired original road network data, Unmatchedtrajectory points are obtained by matching results, Adaptive trajectory clustering is carried out for the unmatched trajectory points, and for each trajectory clustering, the curve fitting of the trajectory points is carried out by using the optimal master curve fitting method, the road centerline is extracted, the driving direction of the road and the single/two-way information are recognized, and then the fusion of the changing road and the original road network is completed. The method can be used to identify the changing area of urban road network quickly, extract and update the fine geometrical structure of the changing road in complex scenario, and identify and update the semantic information of road driving direction, single/two-way and so on.

The invention discloses a road network hot spot region mining method, belongs to the technical field of data mining and solves the problem when trajectory spatial-temporal similarity measurement and clustering calculation is adopted for trajectory clustering in the prior art. The method includes: step 1, performing trajectory segmentation on all trajectory sections, and calculating spatial-temporal similarity and spatial-temporal distance between two sub trajectory sections acquired after segmentation; step 2, performing clustering calculation on all trajectory section data in a grid space according to the spatial-temporal similarity and the spatial-temporal distance of the sub trajectories and a DBSCAN algorithm based on dynamic neighbor; step 3, selecting a remarkable class cluster set from class clusters of clustering calculation, and extracting stay spots from the remarkable class cluster set; step 4, according to the number of trajectory sections carried by the stay spots, acquiring hot regions of the stay spots, and acquiring hot spot regions in a road network from the regions where the stay spots are positioned. The method is used for spatial position locating.

The invention discloses a ship trajectory clustering analysis method based on an improved DBSCAN algorithm. The ship trajectory clustering analysis method comprises the following steps: S1, extractingeffective ship trajectory data from an AIS database; S2, carrying out trajectory similarity measurement by adopting a fusion distance MD, and obtaining a shortest super trajectory fused between trajectories; S3, calculating the length of the shortest super track by adopting the secondary time O (mn); S4, obtaining a fusion distance MD from the length of the track; S5, determining global parameters of the improved DBSCAN algorithm; S6, scanning the whole track segment data set through an improved DBSCAN algorithm to obtain a cluster set; and S7, for the obtained spatial motion mode, if a new track conforms to one of the ship operation modes, considering the track as a normal track. According to the ship trajectory clustering analysis method, the measurement precision is improved; and compared with a traditional DBSCAN algorithm, the time consumption is reduced while a multi-density data set can be well processed, and the robustness and adaptability are good, and tracks are correctly classified.

The invention relates to a road network trajectory clustering-based travel hotspot region mining method. According to the method, taxi trajectories are mapped into a road network, and a clustering method combining points of interest and trajectories collected in actual roads is adopted; on the basis of a density peak clustering algorithm, an OPAM algorithm which optimizes an initial center based on density peaks is provided, namely, a namely DP-OPAM algorithm is provided; according to the algorithm, the local density of data points and shortest distances from the points to points with higher density are adopted, and a decision graph is adopted to select the category of data points with higher density and closest distance, and the category is adopted as an initial clustering center; and onthe basis of the initial clustering center, an OPAM clustering algorithm additionally adopting inverse learning is used to obtain a clustering result. Compared with an original OPAM algorithm, the newalgorithm can not only automatically determine a clustering center, but also improve accuracy and shorten clustering time and realize the analysis of user travel hotspots.

The invention discloses a fast movable object orbit clustering method based on sampling, belonging to the field of orbit data clustering. The method comprises the following steps: dividing each original orbit into an orbit dividing subset by utilizing a minimum length describing principle after a user sets an input parameter and sends a clustering analysis request; carrying out clustering analysis for the orbit dividing subset according to the similarity measurement among line sections to obtain an orbit clustering set; finally generating a representing orbit and a coverage area in a clustering way by each orbit, outputting and visualizing the representing orbit and the coverage area and returning a result to the user. The invention has the advantages that the method is used for the clustering analysis of data of a large-scale low-particlesize movable object orbit and can still keep the orbit clustering effect and also effectively discover the orbit clustering under the condition of very low particlesize of orbit data to be analyzed, thereby enhancing the system efficiency.

The invention discloses a taxi passenger-carrying trajectory clustering algorithm Tr-OPTICS. A research object of the algorithm is passenger-carrying trajectories; reachable distance of trajectories, concepts of core trajectories and search neighborhood scopes of the core trajectories are redefined. For passenger-carrying trajectories of large data volumes, an adjacency list is used for replacing a spatial index in the algorithm, and therefore calculation complexity of the algorithm can be lowered. Algorithm execution efficiency and accuracy of clustering results can be improved via the Tr-OPTICS algorithm put forward in the invention. Stability can be maintained despite different sample sizes, and a frequent pattern of passenger-carrying sub-trajectories can be effectively found based on clustering results via the algorithm disclosed in the invention.

A method, apparatus and computer program product for Functional Space-Time Trajectory Clustering. The method comprises receiving collections of data structures comprising location and time descriptors. The method further comprises estimating functional curves from the collections of data structures. The method further comprises reducing dimensions of the functional curves; and clustering the functional curves into clusters.

The invention provides a method for hierarchical clustering analysis of warship and ship trajectories based on buffer similarity measurement. Similarity measurement on channel trajectory clustering based on buffer analysis is given, and weaknesses of large calculated amount and non-steady clustering effect in a conventional method are overcome; at the same time, situations of a huge amount of warship and ship trajectories and existence of noises are considered, a hierarchical clustering method which is efficient and likely to reduce noises is adopted, and a threshold selection process is revealed. Due to the application of the method, a massive amount of warship and ship trajectories on regional sea can be subjected to hierarchical clustering, so that the distribution of main ship routes is mastered and a foundation for finding activity routines of warships and ships, mentoring marine traffic, improving channel environments, guaranteeing channel safety and the like is laid.

The invention discloses a meta-learning-based vehicle trajectory clustering method and system, belongs to the technical field of vehicle trajectory clustering, and solves the problem that an optimal clustering result cannot be obtained due to the fact that multiple different types of trajectory data adopt a single trajectory clustering algorithm in the prior art. A meta-learning-based vehicle trajectory clustering method comprises the following steps: collecting different types of GPS vehicle trajectory data, and obtaining an optimal DBSCAN clustering algorithm corresponding to the different types of GPS vehicle trajectory data; collecting GPS vehicle trajectory data, and obtaining a meta-learning device used for vehicle trajectory type division through training; and acquiring a vehicle trajectory type corresponding to the GPS vehicle trajectory data by using the meta-learning device, and clustering the GPS vehicle trajectory data by using an optimal DBSCAN clustering algorithm corresponding to the vehicle trajectory type to obtain a clustering result of the GPS vehicle trajectory data. The optimal clustering result can be obtained for various different types of trajectory data.

Embodiments of the invention disclose a travel trajectory clustering method, apparatus and device. The calculation amount of travel trajectory clustering is reduced and the travel trajectory clustering efficiency is improved. The method comprises the steps of obtaining multiple travel trajectories of a user, wherein each travel trajectory comprises a starting point, an ending point and a middle point located between the starting point and the ending point; by utilizing the starting points and/or the ending points of the travel trajectories, clustering the travel trajectories to obtain a first travel trajectory set, wherein the first travel trajectory set comprises the travel trajectories with the matched starting points and/or ending points, and the number of the travel trajectories in the first travel trajectory set is greater than or equal to a first threshold; and by utilizing the middle points in the travel trajectories, clustering the travel trajectories in the first travel trajectory set to obtain a second travel trajectory set, wherein the second travel trajectory set comprises the travel trajectories with the matched starting points and middle points, and/or, the travel trajectories with the matched ending points and middle points.

The invention discloses a moving object trajectory clustering method based on multi-dimensional distance measurement, which comprises the following steps: identifying key points from trajectory data,segmenting original trajectory data, and generating a trajectory segment set; constructing a multi-dimensional distance measurement function based on the spatial distance and the time distance, and calculating the distance between track segments; performing trajectory clustering by adopting a DBSCAN algorithm; and generating a representative trajectory based on a Sweep Line method. According to the method, the key points of the trajectory are identified according to the direction and speed dimension change values of the trajectory data points, so that the redundancy of original trajectory datais reduced, and the trajectory clustering efficiency is improved; and a multi-dimensional distance measurement function is constructed based on the spatial distance and the time distance, so that thetrack clustering precision is improved.

The invention discloses a method of identifying a converse riding behavior of a shared bicycle based on a historical GPS trajectory. A riding behavior converse to a standard motor vehicle driving direction in the direction is defined as a converse riding behavior; through identifying each trip of the shared bicycle and matching the trajectory on a road, abnormal trajectory points are eliminated, converse riding behavior identifying method training is carried out, a trajectory clustering result is used to build a standard trajectory database, a candidate trajectory is subjected to parameter matching, and the converse riding behavior of a shared bicycle is identified. The shared bicycle converse riding behavior identifying reliability and the accuracy are improved, the abnormal trajectory points during the riding process can be quickly identified, a standard trip database can also be used to judge the category of a riding behavior, intervention of the shared bicycle converse riding behavior is thus facilitated, the traffic safety level of shared bicycle riding and motor vehicle driving is improved, and practical application value is reflected.

The invention relates to a crowd sensing Internet of Vehicles user screening method combining clustering and CMAB. The method comprises the following steps: constructing user trajectory clustering features for Internet of Vehicles user driving trajectories; solving an optimal worker combination by using the CMAB model and taking the trajectory clustering information as a basis for user task allocation; and verifying and analyzing the method according to the real taxi track data set. According to the user screening method combining the user trajectory information with the CMAB model, the phenomenon of poor perception data quality caused by low skill level of users participating in perception tasks is overcome, and the selected worker set can have similar driving trajectories; meanwhile, themethod can enable a task allocation algorithm based on a user screening result to tend to be stable faster and earlier. Therefore, the invention is suitable for practical application scenes.

The invention discloses a differential privacy trajectory data protection method based on clustering. The method comprises: adding Laplace noise to count trajectory positions in a class cluster to resist continuous query attacks; secondly, adding radius-limited Laplace noise to the trajectory position data in the class cluster, so that the clustering effect is prevented from being affected by excessive noise; obtaining a noise clustering center of the class cluster according to the noise position data and the noise position count; and finally, defending a non-track position sensitive information attack in the class cluster by utilizing a differential privacy technology. The method has the advantages that the differential privacy technology is applied to trajectory clustering analysis. Forthe position data in each class cluster, Laplace noise is added to the clustering center, and it is avoided that an attacker inquires the specific position data of the user through the adjacent clustering areas. The size of the noise added to the track position is limited, so that the data availability is improved. Laplace noise is added to other information data possibly causing privacy disclosure, and corresponding reasoning association attacks are resisted.

The invention provides an extensible quick trajectory clustering method. Firstly, calculating local MST (minimum spanning tree); secondly, generating a global MST; thirdly extracting a cluster by the method of coarse-grained parallelism or fine-grained parallelism. [Proposed is a new trajectory clustering method based on point data cluster, time expense is lesser than the traditional cluster algorithm based on the model, distance or density, meanwhile, with the stop proposed by the extensible quick trajectory clustering method, the extensibility of trajectory big data clustering is realized.

The invention relates to the technical field of Yangtze River digital navigation channels, and provides a ship route prediction method based on ship trajectory clustering so as to improve the accuracyof ship route prediction. The method comprises the following steps of obtaining the ship trajectory data of a target ship in a navigation channel, and pre-processing the missing values, abnormal values and data formats existing in the ship trajectory data; obtaining a high-quality historical route data set of the ship; clustering the historical routes of the predicted ship, extracting typical characteristic routes of the ship sailing in a control reach, matching the current track of the ship with the characteristic routes, predicting the route selected by the ship for passing through the control reach, and calculating the ship passing time based on the predicted route.

The present invention relates to the field of machine learning, and proposes an ocean trajectory clustering and predicting method. In order to accurately predict future trajectory points, trajectory clustering is required first. According to the trajectory clustering method disclosed by the present invention, similarity measurement is carried out on the trajectory points of complex variability andstrong volatility at sea, and the potential data information is mined; and the method combines the Gaussian mixture model GP with the Dirichlet process DP, and the non-parametric Bayesian framework of the DP is used to determine the number of clusters to improve cluster adaptability. The algorithm uses the process of adding Chinese restaurants based on the DP, and uses the collapsed Gibbs sampling method to solve the model, so that the unsupervised classification from the finite mixed model to the infinite mixed model is implemented, the number of clusters can be automatically obtained, and future trajectory points are predicted for the clustered trajectories by using the Gaussian process regression prediction method. According to the technical scheme of the present invention, the disadvantages of manually specifying the number of clusters and local maximization in parameter estimation are avoided, and the accuracy of prediction is improved under the premise of ensuring adaptive clustering.

The invention relates to a ship AIS trajectory clustering method and device based on a convolution auto-encoder. The ship AIS trajectory clustering method based on a convolutional auto-encoder comprises the following steps: acquiring a continuous trajectory of a ship, and dividing the continuous trajectory into a plurality of sub-trajectories; performing feature engineering extraction on the plurality of sub-trajectories to obtain a sub-trajectory feature matrix; inputting the sub-track feature matrix into a multi-feature fusion auto-encoder to obtain a position feature vector, a speed featurevector and a course feature vector; splicing the position feature vector, the speed feature vector and the course feature vector to obtain a potential feature vector of the ship trajectory; and performing trajectory clustering operation on the extracted ship trajectory feature vector to obtain a ship trajectory clustering result. According to the method, a space-time trajectory measurement methoddoes not need to be selected according to the related data size, trajectory type, calculation complexity, noise and other influence factors, and a similarity distance formula is not needed, so that the calculation time and resources are saved.

The invention relates to a vehicle trajectory clustering method based on graph theory. Firstly, the vehicle networking data is obtained, and the massive vehicle trajectory data is cleaned by using a Spark platform. Secondly, the coordinate points are projected onto the map, and the connected graph is formed according to the set K value and the relative approximation through the relative distance relation of the coordinate points, and then the clusters whose connectivity strength is greater than the preset value are merged by using the interconnectivity; finally, the real taxi trajectory data are analyzed to get the traffic flow of different time periods and different areas, that is, the optimal taxi-hailing scheme is given. The invention can effectively improve the speed and quality of data processing.

The invention discloses a road network trajectory clustering analysis method based on an improved DPC algorithm, and belongs to the field of clustering analysis methods, and the method comprises the steps: S1, carrying out the effective collection of a road moving object trajectory; s2, expressing trajectory data; s3, calculating the distance between the sub-trajectories by adopting an aggregationdistance CD, fusing the vertical distance, the parallel distance and the angle distance by the distance, and obtaining inter-trajectory similarity measurement on the basis of the CD distance; s4, according to the fused CD distance matrix, carrying out sub-track clustering by utilizing an improved DPC algorithm; and S5, for the obtained clustering result, extracting a classical trajectory of the clustering result as a new behavior mode of the whole road network. According to the method, the precision of road trajectory measurement is improved, compared with a traditional density clustering algorithm, the influence of self-defined parameters is reduced, the interactivity with a user is improved, and it is beneficial to discover meaningful trajectory modes in a road network.

The invention relates to a gang mining method, device, equipment and a storage medium. The method comprises the steps of obtaining trajectory data of all objects in a troubleshooting area; performingserialization processing on the trajectory data of all the objects to obtain a trajectory sequence of each object; lCSS-based algorithm, combining the size relationship between the time difference corresponding to the final track points of every two objects and a preset time difference threshold value; calculating the longest common subsequence length between the trajectory sequences of every twoobjects according to the difference between the time corresponding to the starting trajectory point of one object and the time corresponding to the final trajectory point of the other object in everytwo objects and the time difference threshold; and performing clustering analysis according to the longest common subsequence length between the trajectory sequences of every two objects to obtain a trajectory clustering result of the troubleshooting area as a gang mining result. By adopting the method and the system, gang mining can be accurately realized.

The invention discloses a vehicle counting method based on three-dimensional trajectory clustering. The method specifically comprises steps of: acquiring a video image of a road by using a video camera, establishing a relation between a two-dimensional image coordinate and a three-dimensional world coordinate by using a vanishing point method, and acquiring a transfer matrix M; determining a detection line and a detection area in the video image and extracting a background image of the video image; acquiring multiple moving trajectories in the video image; performing coarse clustering on selected moving trajectories satisfying a coarse clustering condition, classifying the selected moving trajectories into multiple categories, and marking the selected moving trajectories as clustered categories; and performing fine clustering on the clustered moving trajectories marked in the step 4. The vehicle counting method is not restricted to environment in engineering application, has high stability and detection precision, is easy to implement and capable of detecting vehicles within the scope accurately in real time, and has wide application prospect.