Aiming at the problem of vehicle objection detection in daily road scenes， a real?time detection method of traffic information based on lightweight YOLOv4 （You Only Look Once version 4） was proposed. Firstly， a multi?scene and multi?period vehicle object dataset was constructed， which was preprocessed by K?means++ algorithm. Secondly， a lightweight YOLOv4 detection model was proposed， in which the backbone network was replaced by MobileNet?v3 to reduce the number of parameters of the model， and the depth separable convolution was introduced to replace the standard convolution in the original network. Finally， combined with label smoothing and annealing cosine algorithms， the activation function Leaky Rectified Linear Unit （LeakyReLU） was used to replace the original activation function in the shallow network of MobileNet?v3 in order to optimize the convergence effect of the model. Experimental results show that the lightweight YOLOv4 has the weight file of 56.4 MB， the detection rate of 85.6 FPS （Frames Per Second）， and the detection precision of 93.35%， verifying that the proposed method can provide the reference for the real?time traffic information detection and its applications in real road scenes.

Face recognition is widely applied in various practical conditions such as entrance guard due to its convenience and practicability. But it is vulnerable to various forms of spoofing attacks (such as photo attacks and video attacks). The liveness detection based on deep Convolution Neural Network (CNN) can solve the above problem but has disadvantages such as high calculation cost, unfriendly interaction mode and difficult deployment on embedded devices. Therefore, a real-time and lightweight security classification method of face recognition was proposed. The face liveness detection algorithm based on color and texture analysis was integrated with the face authentication algorithm, and a face recognition algorithm performing face liveness detection and face authentication in the situation of monocular camera without user cooperation was proposed. The proposed algorithm can support real-time face recognition and has higher liveness recognition rate and robustness. In order to validate the performance of the proposed algorithm, Chinese Academy of Sciences Institute of Automation-Face Anti-Spoofing Dataset (CASIA-FASD) and Replay-Attack dataset were utilized as the benchmark datasets of the experiment. The experimental results show that, in the liveness detection, the proposed algorithm has the Half Total Error Rate (HTER) of 9.7% and Equal Error Rate (EER) of 5.5% respectively, and has the time cost of 0.12 s to process a frame of image in the whole process. The above results verify the feasibility and effectiveness of the proposed algorithm.

Concerning the problem of security for steganography algorithm, a Diamond Encoding (DE) steganography algorithm based on Algebraic MultiGrid (AMG) was proposed. Firstly, an image was divided into two parts of coarse grid and fine grid by the AMG method. Then, the confidential information was embedded into the two part pixels of coarse grid and fine grid by DE method. The change of pixels in coarse grid part has little influence on the whole image quality, while the change of pixels in fine grid part has the great effect on the whole image quality. And the k value of DE is associated with the capacity of information hiding closely, the pixels change greater with the k value increasing. Therefore, in the embedding process with DE, the k value of the coarse grid part is not less than that of the fine grid part. Finally, when the k value of DE was chosen to 1 and 2, three kinds of steganography scheme were proposed. The proposed algorithm was compared with Least Significant Bit (LSB) replacement, random LSB matching, DE algorithm and adaptive edge detection algorithm. The experimental results show that, the first-order Markov security metric of the proposed algorithm is superior to other contrasted steganalysis algorithms.

The existing Eulerian video subtle motion magnification method does not take into account the automatic detection of motion information in the video, the appropriate parameters for motion information processing need to be selected in the realization of motion magnification, such as the filter cut-off frequency, and magnification. For the general video, these parameters usually could not be determined directly, but by trial-and-error. In this paper, an automatic amplification method of non-stationary subtle motion in video was proposed based on S transform. The instantaneous correlation parameters of band-pass filter were automatically determined based on S transform and the corresponding dynamic filter was designed. On this basis, an approach of automated subtle motion magnification was achieved. Firstly, the instantaneous frequency of subtle motion was obtained by S transform in video. Then the dynamic band-pass filter was used to process different frequencies at different times. Finally, the effective motion information which was filtered by band-pass filter was magnified to achieve subtle motion magnification. In addition, for the analysis of anti-noise performance, a method for evaluating the signal-to-noise ratio in video area was proposed. The experimental results show that the proposed method can automatically obtain the parameters such as filter and magnification according to the change of the frequency of the motion information when the actual video is amplified, without manual participation. After the motion amplification, the amplification effect of the moving target can be dynamically displayed. At the same time, the accurate dynamic filter can suppress noise to some extent, and makes the motion magnification effect better.

〖JP2〗With different way of link established, the high frequency IP wide-area network shows a different topological feature during the network operation process. Modeling study and simulation analysis were made with a modified harmonious unifying hybrid preferential model on high frequency IP wide-area network. And the results were found that the degree distribution, shortest path and clustering coefficient would be impacted by the different node selection method for new connection. Mean-while, the average shortest path of the network would be affected by the different edge deletion method. What's more the topological characteristics of the network were determined by the proportion of different node selection method.〖JP〗

According to the emotional states of calm and fear, some emotional information of the Electrocardiogram （ECG）data samples were collected, and researched in both emotional and emotionless states. By the analysis of scatter plots of QT-RR intervals, the 1/f distribution law of ECG signals spectrum and the poincare sections of Heart Rate Variability (HRV), the recognition rate of nonlinear characteristics with these two types of emotional states was calculated, and compared with the recognition rate of statistical characteristics. The results indicate that extracting nonlinear characteristics of ECG signals has a better recognition result compared with statistical characteristics when distinguishing emotional states and emotionless states.

A new adaptive neighborhood Multi-Objective Grid Task Scheduling Algorithm (ANMO-GTSA) was proposed in this paper for the multi-objective job scheduling collaborative optimization problem in grid computing. In the ANMO-GTSA, an adaptive neighborhood method was applied to find the non-inferior set of solutions and maintain the diversity of the multi-objective job scheduling population. The experimental results indicate that the algorithm proposed in this paper can not only balance the multi-objective job scheduling, but also improve the resource utilization and efficiency of task execution. Moreover, the proposed algorithm can achieve better performance on time-dimension and cost-dimension than the traditional Min-min and Max-min algorithms.

To improve the clustering efficiency and image segmentation effect, the paper proposed an Incomplete K-means and Category Optimization (IKCO) method. First of all, the algorithm used simple approach to finish data subsampling and initial centers determining. Then, according to the clustering rules, the proposed algorithm finished image's segmentation. Finally, the algorithm used category optimization method to improve segmentation results. The experimental results show that, compared with the traditional K-means clustering method, the proposed algorithm has better segmentation efficiency, and the segmentation result has a higher consistency with human visual perception.

Based on LEACH, we proposed a novel clustering algorithm Centroidbased Distributed Clustering Scheme（CDCS） for Wireless Sensor Networks（WSNs）. In CDCS, each sensor firstly decided whether it was local tentative clusterheads on its own at any given time with a certain probability popt. The tentative clusterhead computed the centroid of cluster based on information of sensors within cluster; and then dynamically adjusted the structure of cluster, so that the total energy dissipation within the cluster was minimized. Theoretical analysis and simulation results show that CDCS prolong the lifetime of a sensor network by 32%~38% over that of LEACH in different scenes while still maintaining the simplicity of LEACH.

Compelling registration/validation and the ability of anti-cracking are core problems in software copyright protection area. Traditional registration methods mainly depended on the strength and secrecy of validation algorithm, and thus had a low reliability. This paper proposed a new software registration and validation method adopting asymmetric cryptography technique and license file, and also gave the detailed design and implementation of PubCMCenter (Software Publish Control and Manage Center) system based on this method. In the system, the publisher preserves private key, while software users only have the public key that is needed to validate the license file and have no ability to generate the license. Furthermore, this system can also make modular authorization according to the user’s level. Compared with traditional techniques, this system is easier and safer to use.

Bayesian network is a forceful tool to practise inference of uncertainty. It combines graphic theories and probability ones, which can conveniently express and calculate the probability of interesting events and at the same time provide a compact, visual and effective graphic expression for the dependant relationship among the entities. On the basis of testing information independence theory, the test of CI（conditional independence） was carried out on all the joints in the Bayesian network to find out the conditionally dependant relations among them. Then an effective algorithm of Bayesian network structural learning was worked out, which only needed CI testing of O（N ²） times.