To solve the problems of multiple network parameters and high computational complexity in image super-resolution reconstruction of deep network architecture, an image super-resolution reconstruction method based on accelerated residual network was proposed. Firstly, a residual network was constructed to reconstruct the high-frequency residual information between low-resolution image and high-resolution image, so as to reduce the deep network transmission process of redundant information and improve the reconstruction efficiency. Secondly, the dimensionality of the extracted low-resolution feature map was reduced by the feature shrinking layer to realize fast mapping with fewer network parameters. Thirdly, the dimensionality of the high-resolution feature map was increased by the feature expanding layer to reconstruct the high-frequency residual information with the rich information. Finally, the residual and low-resolution images were summed to obtain the reconstructed high-resolution image. Experimental results show that the Peak Signal-to-Noise Ratio (PSNR) and the Structural SIMilarity (SSIM) mean results obtained by the proposed method are 0.57 dB and 0.013 3 higher than those obtained by Super-Resolution using Convolutional Neural Network (SRCNN) respectively, and 0.45 dB and 0.006 7 higher than those obtained by Intermediate Supervision Convolutional Neural Network (ISCNN). In terms of reconstruction speed, using dataset Urban100 as example, the proposed method is 1.5 to 42 times faster than the existing methods. In addition, when this method is applied to the super-resolution reconstruction of motion blur images, it has the performance better than image Super-Resolution using Very Deep convolutional network (VDSR). The proposed method achieves better reconstruction quality with fewer network parameters and provides a new idea for image super-resolution reconstruction.

As an indispensable new resource of modern information society, data always face the risk of being tampered from the beginning. The availability and authenticity of the tampered data will be greatly reduced. And the blockchain technology perfectly meets the requirements of data integrity protection due to its characteristics of anti-tampering, decentralization and single point failure prevention. Firstly, the background of blockchain technology and vital requirements of data protection were briefly described. Secondly, according to the types of blockchain, the existing blockchain data integrity protection achievements were classified and introduced, and the advantages and disadvantages of each achievement were summarized combined with data integrity protection. Thirdly, the current data integrity protection technologies were classified and compared with the blockchain data integrity protection technology, and the shortcomings of the traditional data integrity protection technologies and the advantages of blockchain data integrity protection technology were analyzed. Finally, the defects of blockchain data integrity protection technology were summarized, and the solutions were given.

In recent years, deep learning achieves very good results and has great improvement in object detection. However, in some special scenes, for example, when it is required to simultaneously detect objects with greatly different scales (difference greater than 100 times), common object recognition methods' performance will drop dramatically. Aiming at the problem of recognizing and locating objects with super-large-scale variance in the same scene, the You Only Look Once version3 (YOLOv3) framework was improved, the image pyramid technology was combined to extract the multi-scale features of the image. And in the training process, the strategy of using dynamic Intersection over Union (IoU) was proposed for different scale objects, which was able to better solve the problem of sample imbalance. Experimental results show that the proposed model significantly improves the recognition ability of super-large and super-small objects in the same scene. The proposed model has been applied to the airport environment and achieved good application results.

The tiny lesion area and the large difference between samples of neonatal punctate white matter lesion make it difficult to detect and segment the lesion. To solve the problem, a refined deep residual U-Net was proposed to realize the fine semantic segment of the lesion. Firstly, a Magnetic Resonance Imaging (MRI) image was cut into small patches. Secondly, the deep features of multiple layers of each image patch were extracted by the residual U-Net. Then, the features were fused and the probability map of the lesion distribution of each image patch was obtained. Finally, the probability map after splicing was optimized by the fully-connected condition random field to obtain the final segmentation results. The performance of the algorithm was evaluated on a dataset provided by a cooperative hospital. The results show that with only T1 order unimodal data used, the proposed model has the lesion's edge segmented more precisely, and the anti-interference ability of the model is prominent. The model has the Dice similarity coefficient of 62.51%, the sensitivity of 69.76%, the specificity of 99.96%, and the modified Hausdorff distance reduced to 33.67.

To solve the limitation and accuracy of matrix decomposition in Collaborative Filtering (CF) algorithm, a Collaborative Filtering algorithm based on Bounded Matrix low rank Approximation (BMA) and Nearest neighbor model (BMAN-CF) was proposed to improve the accuracy of item scoring prediction. Firstly, the matrix factorization algorithm of BMA was introduced to extract the implicit feature information of sub-matrix and improve the accuracy of neighborhood set search. Then, the target users' scores on target items were respectively predicted according to the traditional user-based and item-based collaborative filtering algorithms. And the equilibrium factor and control factor were used to dynamically balance the two prediction results, the target users' scores of items were obtained. Finally, the data was partitioned, and the proposed algorithm was parallelized in Hadoop environment by using the characteristics of MapReduce computing framework. The experimental results show that, the BMAN-CF has higher rating prediction accuracy than other matrix factorization algorithms, and the speedup experiment shows that the proposed parallelized algorithm has better scalability.

Concerning DOM XSS (Document Object Model (DOM)-based Cross Site Scripting (XSS)) vulnerability detection in Web client, a detection algorithm for DOM XSS vulnerability based on dynamic taint analysis was proposed. By constructing DOM model and modifying Firefox SpiderMonkey script engine, a dynamic taint analysis method based on the bytecode was used to detect DOM XSS vulnerabilities. First, taint data was marked by extending the attribute of the DOM object class and modifying the string encoding format of SpiderMonkey. Then, the execution route of the bytecode was traversed to generate the tainted data set. After that, all the output points which might trigger DOM XSS attacks were monitored to determine whether the application contained the DOM XSS vulnerabilities. In the experiment, a DOM XSS vulnerability detection system containing a crawler was designed and implemented. The experimental results show that the proposed algorithm can effectively detect the DOM XSS vulnerabilities, and the detection rate is about 92%.

Focused on the limitation of monitoring distance, restriction of online transmission, large amount of information and that high-power data-link is disable to board small environment monitoring Unmanned Aerial Vehicle (UAV), a multiple-UAV environment monitoring task scheduling method considering 3G/4G network features was proposed. First, the time characteristic of 3G/4G network and the multiple-UAV environment monitoring task scheduling were combined, and this issue was modeled as Team Orienteering Problem with Time Window (TOPTW). Secondly, since the problem of huge computation and easily falling into local optimum, an Iterated Local Search (ILS) algorithm was proposed to get the optimization solution. Thirdly, a large amount of test data sets were applied into experiments to verify the feasibility and computing performance, and the comparative result between ILS and Ant Colony Algorithm (ACA) about the average profit and computing time were proposed. Last, the algorithm was applied in typical two UAV environment monitoring task scheduling under 3G/4G network. The results show that, the most profits received from ILS were worse than those from ACA. The average Gap of all test data sets was 1.09% and the largest was 10.8%. There were some results better than those in ACA. And the computing time of ILS was nearly reduced to a thousandth of the computing time of ACA. The experimental results show that ILS algorithm can fast solve the issue of multi-UAV environment monitoring task scheduling and effectively reduce the computing time with profit results in an acceptable scope.

The inconsistence of normal curvature at vertex when reconstructing a coarse mesh model as a fine heavy one has been unsolved and this inconsistence will result in shadow in rendering. In this paper, the geometric condition of the normal curvature consistence at vertex was obtained and a novel algorithm was further proposed based on that condition refining coarse mesh models to be visually smooth parametric ones, which were presented collectedly as triangular Gregory patches. The constructed parametric model was G ¹ everywhere without the normal curvature inconsistence problem, so a good visual effect could be obtained. The experimental results show that the proposed algorithm can obtain a high quality visual effect with the 1%-2% vertexes of the original mesh model.

In task decomposition and coordination for multi-robot, the problem of workload imbalance in task with partial order constraint still exists. To overcome this problem, a task coordination and workload balance method for multi-robot based on trading-tree was proposed. Firstly, the task decomposition problem satisfying partial order constraint was described as a constraint graph. Secondly, an initial task assignment strategy was proposed according to the directed weighted graph, and the problem of task coordination among multiple robots was solved by using the improved Dijkstra algorithm. At last, the strategy of workload balance was proposed to balance each robot's workload without violating any constraints via a protocol based on trading-tree. The experimental results show that, compared with Dijkstra algorithm, after finishing the strategy of workload balance, the efficiency signifiantly increases by 12% and the difference of workload reduces by 30%.

To deal with the time-varying and nonlinear properties of network security and its difficulty in prediction assessment, a network security situation prediction method based on Harmony Search algorithm and Relevance Vector Machine (HS-RVM) was proposed to offset the prediction accuracy drawbacks of existing prediction methods. In the prediction process, network security situation samples were firstly normalized and phase space was reconstructed; then, Harmony Search (HS) algorithm was adopted to find out the optimal Relevance Vector Machine (RVM) hyper parameters to build the network security situation prediction model with improved prediction accuracy and velocity; finally, Wilcoxon signed rank tests were used to testify the difference of prediction performance. The simulation cases indicate that the Mean Absolute Percentage Error (MAPE) and the Root-Mean-Square Error (RMSE) of the proposed prediction method are 0.49575 and 0.02096 respectively, with a better prediction performance than the Improved Harmony Search (IHS) algorithm and Regularized Extreme Learning Machine (IHS-RELM) prediction model and PSO and Support Vector machine for Regression (PSO-SVR) prediction model. The outcome of Wilcoxon signed rank tests show there is a significant difference. The proposed method is capable to depict the changing rules of network security situation relatively, which is helpful for network administrators to control the changing tendency of network security situation in time.

Focusing on the issue that the tracking algorithm based on template matching has poor performance in running speed and success rate, a template matching tracking algorithm based on Particle Swarm Optimization (PSO) was proposed. The algorithm took the PSO algorithm as the search strategy of the candidate templates in template matching algorithm, and the target template was updated self-adaptively. Firstly, 30 candidate templates were selected in a search scope and then the individual and global optimal candidate template were selected; secondly, the best candidate template was worked out through the particle swarm optimization and the target is the best one; finally, the target template was updated self-adaptively based on the matching rate of the best candidate template. The theoretical analysis and simulation experiments show that, compared with the tracking algorithm based on template matching and the template matching tracking algorithm based on the rough search and refined by search, the computation of the template matching tracking algorithm based on particle swarm optimization is greatly reduced about 91.1% and 69.8%, and the success rate is 2.02 times and 1.94 times of the primary algorithm. The experiment show that the new algorithm can achieve well real-time tracking and the robustness and accuracy of tracking is greatly improved.

To solve the attitude drift, low real-time ability and high price problem in motion capture system based on inertial sensors, a kind of real-time motion capture system was designed to effectively overcome the attitude drift with low cost and power consumption. At first, a distributed joint motion capture node was built based on the human body kinematics principle, and every node worked in low-power mode, when the acquisition data from the node was lower than a predetermined threshold, the node would automatically enter into the sleep mode to reduce the power consumption of the system. In order to reduce the data drift in traditional algorithm, a kind of algorithm combined with inertial navigation and Kalman filter algorithm was designed to calculate the real-time motion data. Using the Wi-Fi module, the TCP-IP protocol was adopted to transmit the attitude data, which could drive the model in real time. At last, the accuracy of the algorithm was evaluated on the multi-axis motor test platform, and the effect of the system for tracking real human motion was compared. The experimental results show that the algorithm has higher accuracy by contrast with the traditional complementary filtering algorithm, which can control the angle drift in less than one degree; and the delay has no obvious lag by contrast with the complementary filter, which can realize the accurate tracking of human motion.

In order to solve the problems in Chinese sentence grouping domain, including the lack of computational linguistics data and the joint makers in a discourse, this paper proposed an automatic Chinese sentence grouping method based on Multiple Discriminant Analysis (MDA). Moreover, sentences group was rarely considered as a grammar unit. An annotated evaluation corpus for Chinese sentence group was constructed based on Chinese sentence group theory. And then, a group of evaluation functions J was designed based on the MDA method to realize automatic Chinese sentence grouping. The experimental results show that the length of a segmented unit and one discourse's joint makers contribute to the performance of Chinese sentence group. And the Skip-Gram model has a better effect than the traditional Vector Space Model (VSM). The evaluation parameter P_μ reaches to 85.37% and WindowDiff reduces to 24.08% respectively. The proposed method has better grouping performance than that of the original MDA method.

Given the feature extraction of the furnace flame image produced in boilers and industrial production, a hierarchical adaptive method to extract salient points was proposed. First the Block Difference of Inverse Probabilities (BDIP) model was used to change the original image into BDIP image. On the basis of this, the BDIP image was made into Haar wavelet transform, the salient value of two-dimensional image was calculated by the improved weighted method, and then a non-equilibrium quadtree was built through the proposed adaptive method. The root of quadtree represented the salient value of the image, and the salient points number of subtree was determined according to the ratio of the salient value of every subtree to the salient value of parent node. The proposed extracting algorithm was salient points compared with the extracting algorithms based on BDIP and based on Haar wavelet transform. The experimental results show that edge accuracy and comprehensive feature retrieval accuracy at least increase by 10% and 3.5% respectively. The proposed method overcomes the shortcoming of traditional way that it extracts too many salient points and some extracted points are not salient, at the same time the method avoids local gather of salient points.

The existing Identity-Based Online/Offline Encryption (IBOOE) schemes do not allow the attacker to choose the target identity adaptively, since they are only proven to be secure in the selective model. This paper introduced the online/offline technology into fully secure Identity-Based Encryption (IBE) schemes, and proposed a fully secure IBOOE scheme. Based on three static assumptions in composite order groups, this scheme was proven to be fully secure with the dual system encryption methodology. Compared with the famous IBOOE schemes, the proposed scheme not only greatly improves the efficiency of the online encryption, but also can meet the demands for complete safety in the practical systems.

To solve the problem of detecting Denial of Service (DoS) and network probing attacks, a new method based on Principal Component Analysis (PCA) was proposed in this paper. PCA was done on both attack and normal traffic to collect various statistics, and then the detection model was constructed based on these statistics. At last, this paper utilized the threshold of the statistics to achieve a fixed rate of false alarms. The experimental results show that this approach can detect DoS and network probing attacks effectively, and yield 99 percent detection rate; in addition, security masters can make responses in time and the responses can reduce the loss under real-time attacks.

Given the improper threshold, Sobel operator could easily lead to the loss of image edges or produce broad pseudo-edges. To solve these problems, appropriate threshold was selected by using the constraint of variance firstly, and then mathematical morphology thinning algorithm was utilized to thin the edge images detected by Sobel operator. The experimental results show that this method can bring about satisfactory thinning effects while retaining the original edge information.

It is indicated that Cao Guoliang et al's multicast key management scheme based on generalized cat map does not satisfy the independent security requirement of individual keys of group members. A modification scheme was proposed by using one way function. The presented scheme satisfies the independent security requirement of individual keys of group members. Compared with Cao Guoliang et al's scheme, this modification scheme has higher security and lower computation and communication overheads.

The limited node energy, high node redundancy and other characteristics of Wireless Sensor Network (WSN) make working-in-round mechanism one of the basic policies in solving the network coverage problem. In this paper, the authors investigated the network characteristics of stochastic scheduling model, analyzed the relationship between the number of effective nodes and rotation cycles, and finally provided an adaptive algorithm to adjust the node work probability according to the effective node number in the network. The algorithm can solve the performance problem in the later periods of the network, which is caused by reduction of effective nodes and static work probability, and thus guaranteeing the network performance in each round. Simulation verifies the effectiveness and correctness of the novel algorithm.