In order to achieve the precise semantic correlation between image and text， an image text retrieval method based on Feature Enhancement and Semantic Correlation Matching （FESCM） was proposed. Firstly， through the feature enhancement representation module， the multi-head self-attention mechanism was introduced to enhance image region features and text word features to reduce the interference of redundant information to alignment of image region and text word. Secondly， the semantic correlation matching module was used to not only capture the corresponding correlation between locally significant objects by local matching， but also incorporate the image background information into the global image features and achieve accurate global semantic correlation by global matching. Finally， the local matching scores and global matching scores were used to obtain the final matching scores of images and texts. The experimental results show that the FESCM-based image text retrieval method improves the recall sum over the extended visual semantic embedding method by 5.7 and 7.5 percentage points on Flickr8k and Flickr30k benchmark datasets， respectively； the recall sum is improved by 3.7 percentage points over the Two-Stream Hierarchical Similarity Reasoning method on the MS-COCO dataset. The proposed method can effectively improve the accuracy of image text retrieval and realize the semantic connection between image and text.

Aiming at the problems of low sensitivity and high false positive rate caused by various shapes and difficulty in detecting pulmonary nodules by the Computer-Aided Detection （CAD） system of pulmonary nodules， a pulmonary nodule detection algorithm based on attention feature pyramid networks was proposed. In the first stage， a more compact Dual Path Network （DPN） was used as the backbone network， and a Feature Pyramid Network （FPN） was combined for multi-scale prediction to obtain feature information at different levels. At the same time， the Global Attention Mechanism （GAM） was embedded to refine the semantic features to be emphasized in learning and improve the sensitivity of the algorithm. In the second stage， a false positive reduction network was proposed to obtain the final classification prediction results. In the training stage， the focal loss function and various data augmentation techniques were used to deal with the data imbalance problem. Experimental results on the public dataset LUNA16 （LUng Nodule Analysis 2016） show that the Competitive Performance Metric （CPM） of the algorithm only with the first stage reaches 0.908， and after adding the false positive reduction network， the CPM of the algorithm reaches 0.933， which is 1.1 percentage points higher than that of the classic algorithm — Convolutional Neural Network （CNN） based on Maximum Intensity Projection （MIP）. And ablation experimental results show that the dual path network， FPN， and GAM are effective in improving the detection sensitivity. The above proves that the proposed two-stage detection algorithm can obtain multi-scale nodule information， improve the sensitivity of pulmonary nodule detection， and reduce the false positive rate.

The pictures taken on hazy days have color distortion and blurry details， which will affect the quality of the pictures to a certain extent. Many deep learning based methods have good results on synthetic homogeneous haze images， but they have poor results on the real nonhomogeneous dehazing dataset introduced in the latest NTIRE （New Trends in Image Restoration and Enhancement） challenge. The main reason is that the non-uniform distribution of haze is complicated， and the texture details are easily lost in the process of dehazing. Moreover， the sample number of this dataset is limited， which is easy to lead to overfitting. Therefore， a Conditional Generative Adversarial Network with Dual-Branch generators （DB-CGAN） was proposed. Among them， in one branch， with U-net used as the basic architecture， through the strategy of "Strengthen-Operate-Subtract"， enhancement modules were added to the decoder to enhance the recovery of features in the decoder， and the dense feature fusion was used to build enough connections for non-adjacent levels. In the other branch， a multi-layer residual structure was used to speed up the training of the network， and a large number of channel attention modules were concatenated to extract more high-frequency detailed features as many as possible. Finally， a simple and efficient fusion subnet was used to fuse the two branches. In the experiment， this model is significantly better than the previous Dark Channel Prior （DCP）， All-in-One Dehazing Network （AODNet）， Gated Context Aggregation Network （GCANet）， and Multi-Scale Boosted Dehazing Network （MSBDN） dehazing models in the evaluation index Peak Signal-to-Noise Ratio （PSNR） and Structural SIMilarity （SSIM）. Experimental results show that the proposed network has better performance on nonhomogeneous dehazing datasets.

Forged and tampered data frames should be identified and filtered out to ensure network security and efficiency. However， the existing schemes usually fail to work when verification devices are attacked or maliciously controlled in the Software Defined Network （SDN）. To solve the above problem， a blockchain-based data frame security verification mechanism was proposed. Firstly， a Proof of Frame Forwarding （PoFF） consensus algorithm was designed and used to build a lightweight blockchain system. Then， an efficient data frame security verifying scheme for SDN data frame was proposed on the basis of this blockchain system. Finally， a flexible semi-random verifying scheme was presented to balance the verification efficiency and the resource cost. Simulation results show that compared with the hash chain based verifying scheme， the proposed scheme decreases the missed detection rate significantly when an equal proportion of switches are maliciously controlled. Specifically， when the proportion is 40%， the decrease effect is very obvious， the missed detection rate can still be kept no more than 32% in the basic verification mode， and can be further reduced to 7% with the assistance of the semi-random verifying scheme. Both are much lower than the missed detection rate of 72% in the hash chain based verifying scheme， and the resource overhead and communication cost introduced by the proposed mechanism are within a reasonable range. Additionally， the proposed scheme can still maintain good verification performance and efficiency even when the SDN controller is completely unable to work.

In the simulation of Software Defined Network (SDN), the existing network simulation tools usually do not consider the processing delay of SDN switchs. To make the simulation result more realistic and accurate, a scheme to simulate the processing delay was proposed. First, the scheme divided the process of the switch forwarding into two aspects: inquiry operations on flow table and execution of various actions, and then transferred the two aspects into processing delay by using processor frequency and memory cycle. Measurement and comparison were conducted on the processing delay of switches with different configuration in real and simulation environments. The results show that the simulated processing delay of the proposed method is almost close to that in real environment, it can accurately estimate the processing delay of switches.

Language transmission network is a typical social network, the structure and dynamics of language networks have a significant impact on competition and the spread of the language. Therefore, using the language competition in the same area as the object of study, and then the paper proposed an Agent-based social circles network to build the social network closer to the actual language. The whole network parameters and structural parameters of individual networks have good social network characteristics. Agents in the network could be distributed to a social circles of different size. They can move, born and die, which led to the disconnection of previous links and the establishment of new contacts. Each Agent adopted one of three possible states: monolingual language in X, monolingual language in Y and bilingual language in Z, and transmitted horizontally and vertically. On the basis of the analysis of the language status, attractive parameter, the peak rate of horizontal and vertical transmission, the proportion of speakers on the impact of language competition, the article analyzed the impact of social interaction radius and social mobility on language competition. The simulation results indicate that compared with the static social network model, the proposed model is closer to the actual society, and it can effectively increase the likelihood of coexistence between languages, provides a better environment for the study of the preservation of endangered language maintenance.

Concerning the random prediction results of the neural network model, a compact wavelet neural network was constructed. The method transferred the wavelet function into the hidden layer of the Back-Propagation (BP) network and made use of a random certain state command to obtain the definite prediction results. Compared with the wavelet neural network realized by programming and BP network, this method is suitable for mass data training and has such advantages as strong adaptability and robustness for data samples, especially has better adaptability for high frequency stochastic time series, and has characteristics of determined predicted results, powerful practicability and so on. It can obviously improve the training speed, prediction accuracy and prediction efficiency of the model. Its efficiency has been proved by the gas emission prediction experiment of wavelet packet transformation and wavelet neural network.

To maximize the lifetime of Wireless Sensor Network (WSN) with small link load and less network delay, the Iterative Geometric Programming (IGP) algorithm of joint routing and sleep scheduling was analyzed and researched, and an improved algorithm was proposed. The improved algorithm counted the packets sent and received by the node and the number of idle cycles for a period of time. According to the record, sleep time which made the node work power smallest was calculated, and then the time was set as the node sleep time for next period. Last, the work power was transmitted to its adjacent nodes and the node residual energy was forecasted. Therefore, the energy route selection was done. The experimental results show that the improved algorithm prolongs the network lifetime about 23% and reduces network delay.

The three-dimensional spatial target localization based on wireless communication and networking technology is a hot research topic in the field of Internet of Things.However,there are still some problems including location is not accurate enough,the calculation overhead is too high and power consumption is too large.Thus,a distributed three-dimensional localization mechanism for the environment of the Internet of Things is proposed.The algorithm use Cooperative Location-Sensing(CLS) to mesh grid and determine the target location by the estimated distance.It combines Gaussian fitting ,signal sorting mechanism and Bounding-inbox algorithm which have effectively reduced signal interference.Moveover,it use local meshing to reduce the grid voting overhead.Simulation results show that the algorithm is better than the existing three-dimensional localization algorithm in positioning accuracy and have a lower power consumption.

In order to seek for applications in video coding of Compressed Sensing (CS) and improve the coding efficiency of MPEG-2, a CS and MPEG-2 based improved scheme was proposed. The improved video coding scheme chose the method producing an image with smaller Sum of Squared Differences (SSD) as the final reconstruction method between the standard reconstruction method and the Total Variation (TV) minimization algorithm in the pixel domain, which is based on the fact that the original image has sparser gradient than the residual image. The experimental results show that the proposed scheme is efficient for all kinds of video sequences. The improvement of Peak Signal-to-Noise Ratio (PSNR) is greater than 0.5dB for the sequences with sharp edges, and 0.26dB~0.41dB for sequences with smooth areas or complex textures.

A median priori image reconstruction algorithm based on mixed model was put forward to solve the problems of over-smoothness and stepladder edge of reconstructed image by Maximum A Posterior (MAP). First,in the median priori distribution of MAP reconstruction method,the combination of wavelet shrinkage and forward-and-backward anisotropic diffusion filter was introduced before each of median filtering. In addition, if the background area still kept a small amount of noise, the fine filter with a nonlinear diffusion that smoothed the smaller image gradient threshold region could be chosen to join in the last of iteration,so as to optimize the image.The simulation results show that the algorithm has good performance in both lowering noise effect and preserving edges. Compared with other classical algorithms,the Signal-to-Noise Ratio (SNR) can be improved by 0.9dB to 3.8dB.

The existence of potential links in sparse networks becomes a big challenge for link prediction. The paper introduced active learning into the link prediction task in order to mine the potential information of a large number of unconnected node pairs in networks. The most uncertain ones of the unlabeled examples to the system were selected and then labeled by the users. These examples would give the system a higher information gain. The experimental results in a real communication network dataset Nodobo show that the proposed method using active learning improves the accuracy of predicting potential contacts for communication users.

By introducing an indicator to evaluate performance of Multidimensional Independent Component Analysis (MICA) algorithm, the separation was studied by numerical simulation. The multidimensional Amari separation error was used as an important indicator of the measurement of MICA algorithm performance. In the comparative separation performance analysis of four algorithms named vkMICA, cfMICA, MSOBI, SJADE, a random distribution of letters signal was used for simulation and testing, and a visual representation of MICA model of separation and uncertainty was got. The results show that MICA is a very effective method for multidimensional source signal analysis.

Determination of index weights in the current heterogeneous network selection algorithm is of subjectivity. To solve this, a network selection algorithm based on fuzzy Analytic Hierarchy Process (AHP) and extension theory was proposed. Based on the analysis of the network requirements of different business, and the combination of extension theory, the collected performance parameters were mapped into corresponding calibration interval. Through the establishment of the network and calculation of the relative membership degree of the matter-element, a new decision matrix was constructed. The integrated weights of the network performance parameters were calculated by fuzzy AHP. Finally, the optimal access network was chosen through the weighted ranking of the relative membership degree. The simulation results show that the algorithm can take account of the business types of different users' and network objective performance, and multi-mode terminal can select the network accurately and effectively in the heterogeneous networks.

All the available geometry theorem provers are not sustainable. A knowledge representation with the general structure and a reasoning algorithm which could deal with all the rules were proposed. According to these ideas, a geometry automatic reasoning platform that could be sustainably developed by the user had been initially implemented. This platform allows the user to add geometric knowledge such as geometric objects, predicates and rules, and provides multiple reasoning algorithms such as forward search method and a part of area method, so it will be more suitable for geometry teaching.

This paper proposed an earthquake prediction method based on neural networks and an improved particle swarm optimization algorithm. In this method, a feed forward neural network was applied to predict the level of earthquake, and a modified particle swarm optimization algorithm was applied to optimize the neural network model. In order to get weights of the optimal balance between the global search and local search, a Dynamic Mutational Particle Swarm Optimization (DMPSO) algorithm was designed by using the ideology of dynamic mutation. This algorithm was used to adjust weights of the feed forward neural network. The simulation results of the proposed method were compared with the simulation results of two feed forward networks with different training algorithms. The comparison results show that the prediction model with DMPSO has fastest convergence rate, the smallest prediction error and strongest generalization ability. In conclusion, the model with DMPSO is a good reference to the middle earthquake prediction.

An AdaBoostSVM (AdaBoost Support Vector Machine) algorithm applied to content-based image retrieval was proposed. It uses Support Vector Machine (SVM) as component classifier of the AdaBoost algorithm, and simulates the basic sample re-weighting method of AdaBoost algorithm by adding important samples based on relevance feedback mechanism. The experimental results show that the AdaBoostSVM algorithm can improve the performance of retrieval system in the database of 2000 images effectively.

The multi-designated verifiers signature is a special digital signature that the signature could only be checked by several designated verifiers. A multi-designated verified signature scheme and an identity based multi-designated verified signature scheme were proposed based on the technique of multilinear forms. The security analysis of the schemes shows that they have the property of unforgeability, source hiding and privacy of signer's identity.

Authentication is an issue of importance in computer communications, and password authentication protocols have been widely utilized due to their great convenience. Recently, Lee et al. proposed a nonce-based remote user authentication scheme using smart cards. In this paper, however, it is found that their scheme is not secure as claimed, and two attacks can be launched to break the scheme.

Considering the characteristics of sensor nodes' limited energy and limited transmission radius, the clustering hierarchy tree routing algorithm based on LEACH was proposed. In cluster construction stage, the optimal candidate cluster heads would become the final cluster heads, which guaranteed the distance between any two heads was far enough and neither of them was in the same area of radius, so the cluster heads were uniformly distributed in the network. In the intercluster communication, cluster heads form a hierarchical tree and the root was Base Station (BS), which reduced the energy consumption of direct data transmission from the head to BS, thereby prolonging the life cycle of the network.