According to the passenger movement pattern and the hot pick-up and drop-off areas extracted from taxi driving passenger data,this paper proposes a functions discovery method of taxi hot areas.Firstly,it uses taxi driving data clustering algorithm based on the temporal and spatial characteristics of traffic data to realize hot region division.Then,the passengers travel character discovery model of passengers in hot region based on Latent Dirichlet Allocation(LDA) is built to realize clustering hot taxi region with similar passenger travel mode.Finally,by summarizing the specific function of each area,it can find the relationship between area function and passenger movement patterns at different period of passenger flow.The experimental results show the method can effectively discovery the function characteristics of hot areas.
Aiming at the problems in traffic sign detection process,such as that traffic sign shows dimensional changes,rotation distortion,projection distortion or the sign is partially occluded,in this paper,a traffic sign detection algorithm based on saliency map and Fourier Descriptor(FD) is proposed.Firstly,the Frequency Tuning(FT) method is used to get the saliency map.Secondly,binary operation is utilized on saliency map to achieve binary image and get the regions of traffic sign.By extracting outer contour of the regions,features of contour length and aspect ratio are utilized to filter interference information.Thirdly,the convex hull processing is applied for the eligible contours and FDs of convex hull are extracted and normalized.Finally,detection results are obtained according to the comparison with standard data.Experimental results demonstrate that the detection rate of the proposed algorithm is more than 95%.This algorithm meets the real-time performance requirement of traffic sign detection.
In the rail transit clusters dispatch system,the parameters of database connection pool is one-time set and can not be modified.Aiming at this shortage,a dynamic allocation strategy of multi-client database connection pool is designed.It uses a dynamic allocation algorithm to allocate the optimal number of connections according to the different access frequency of each client for current client,so as to achieve the purpose of improving the utilization of system resources.Experimental results show that the dynamic allocation strategy proposed in this paper can shorten the response time of connection pool and improve the efficiency of system.
Currently HTTPS protocolused by Web applications provides a good message encryption mechanism for the message requests between the client and the server,but it cannot selectively process the application layer message,thus the whole process message request interaction does not encrypt,which cannot prevent replay attacks in the whole process.So,this paper proposes an anti-replay attack scheme based on counterand dynamic calibration,which uses the current time on the server as the dynamic calibration,and sets the counter mechanisms feedback client data packet loss.Experimental results show the scheme compared with the sheme based on the single serial number mechanism can relieve dependence on the database server.Compared with thesheme based on a single time stamp mechanism,it can avoid clock synchronization and feedback data packet loss.
On the basis of Histogram of Oriented Gradient with Support Vector Machine(HOG-SVM) algorithm and LeNet network model, a pedestrian detection algorithm which is the combination of HOG and Convolutional Neural Network(CNN) is proposed. Firstly, multi-scale sliding windows are used to extract the HOG features which are then sent to SVM classifier to find the candidate regions.The regions are judged according to the posterior probability. And the CNN algorithm is used to eliminate the false detection window. In order to solve the problem that a single target is framed by multiple candidate regions, the Non-maximum Suppression(NMS) algorithm is used to fuse the multi-rectangles, remaining the largest posterior probabilitywindow and suppressing the overlapped windows.In the classifying process, the candidate region is judged as pedestrian region based on the posterior probability in SVM and LeNet. Experimental results show that this algorithm can get higher recognition rate and recall rate compared with HOG-SVM and LeNet algorithms.
Aiming at the theme drifting and the page weight splitting of traditional PageRank algorithm,an improved PageRank algorithm is proposed.In order to improve the user query efficiency and search quality,combined with the time feedback factor,it makes a comprehensive analysis on user forwarding,user comments and micro-blog mentions.Statistical analysis is used to measure the contribution of user behavior in the ranking of micro-blog user influence.By using the improved TF-IDF algorithm to calculate the similarity weight of the topic,the user can select the Web page with high relevance to obtain the corresponding PageRank weight.Experimental results show that compared with common microblog ranking algorithms,the improved PageRank algorithm has better user influence ranking effect.
In order to enhance the robustness and the instantaneity of the traditional vehicle recognition algorithms,a fine vehicle make and model recognition algorithm is proposed by incorporating Locality-constrained Linear Coding(LLC) and the weighted Spatial Pyramid Matching(SPM) model.Firstly,the Histogram of Oriented Gradient(HOG) feature is extracted to obtain the image representation vector which improves the recognition accuracy.Then the weighted SPM is applied to integrate the spatial information of vehicle images into their final representation vector.Finally,the final representation of vehicle images generated by above steps is sent to a linear Support Vector Machine(SVM) classifier for training and testing.Experiments are conducted on 56 827 vehicle images coming from 150 vehicle makes and models which are captured by real-time traffic surveillance system in various weather conditions and illuminations.The results demonstrate that the proposed algorithm has better performance both in recognition accuracy and the recognition time.
Aiming at the problem that the interpolation point gray estimation is not accurate in the process of image reconstruction,a single-image super-resolution algorithm in frequency domain based on neighborhood feature learning is proposed in this paper.When giving a low-resolution image as input,it uses image feature to learn local covariance from low-resolution image and its corresponding high-resolution image’s geometric similar structure.For each patch in the neighborhood,four directional variances are estimated to adapt the interpolated pixels.Experimental results demonstrate that the proposed method not only can guarantee the consistency of the smooth region in the reconstructed high-resolution image,but also can retain the image details and the integrity of the edge profile.
Current researches concerning network rumor control strategy mainly focus on the effect of high connectivity on information propagation,namely,the number of neighbor nodes,and pay less attention to the indirect effects induced by neighbor information.Therefore,this paper proposes a Self-degree and Neighbor-degree(SDND) rumor immunization strategy,which merely needs to know the local information of network.In this strategy,immunization node is selected by comprehensively considering on the out-degree of the node and the maximum out-degree of the neighbor nodes.Subsequently,based on the dataset of Sina microblog,the effects of target immunization,acquaintance immunization,important acquaintance immunization,and SDND immunization on rumor propagation are compared by simulating rumor propagation using the SEIR rumor propagation model.Simulation results show that SDND presents a superior immune effect than target immunization and acquaintance immunization strategies.It can effectively inhibit rumor propagation.
This paper proposes a computable filling method,aiming at the problem of deformable filling under certain shape constraint conditions.Quadrilateral mesh split of the target area and filling template is done.Integer planning is presented under constraint conditions such as template connecting,boundary,rotation,minimum deformation,etc.Discrete splicing in filling area is realized using filling template.Through the global optimization,model deformation is iterated,so as to achieve the desired filling effect.Experimental results show that the proposed filling method has no direct relationship with the constraints in domain coverage ratio of target area and edge fitting degree,and it can better achieve region filling under the specified constraints.