Concerning of the security and efficiency problems of the agricultural product traceability system, based on the decentralization security feature of blockchain, an efficient solution based on consortium blockchain was proposed. Firstly, through Inter-Planetary File System (IPFS), the agricultural product data was hashed, so as to reduce the data size of single transactions in the block, and the initial guarantee of data was achieved by using the irreversible principle of IPFS data. Secondly, the consortium blockchain model for data verification was established, and Practical Byzantine Fault Tolerant (PBFT) algorithm was used as consensus algorithm for blockchain data verification to reduce the consensus time of the whole network. Finally, according to the number of participating nodes, block size and network bandwidth in the simulation experiment, the time curve of the verification transaction was fitted, and then the blockchain transaction efficiency under different bandwidths was calculated; by using tens of thousands of actual situations of the agricultural product traceability system with the participation of sensors, the blockchain double-chain structure was compared to obtain the analysis results. Experimental results show that under the condition of less than 1000 verification nodes, the maximum consensus time of blockchain is 32 min, and the consortium blockchain system can support 350000-400000 sensor data, which can be applied to large-scale and multi-data agricultural product traceability.

To solve the problems of the cumbersome process in medical record authorization, the low efficiency in record sharing and the difficulty in identity authentication in current medical systems, a method of asymmetric encryption technology combining with blockchain technology was proposed to make medical records cross-domain sharing traceable, data tamper-resistant and identity authentication simplified by applying charatistics of asymmetric encryption technology like high safety and simple cooperation to the peer-to-peer network constructed by blockchain technology. Firstly, based on the anti-tampering of blockchain technology and with asymmetric encryption technology combined, file synchronization contract and authorization contract were designed, in which the distributed storage advantages secure the privacy of user's medical information. Secondly, cross-domain acquisition contracts were designed to validate the identity of both parties and improve authentication efficiency, so that non-legitimate users can be securely filtered without third-party notary agency. The experimental and analysis results show that the proposed scheme has obvious advantages in data guard against theft, multi-party authentication and data access control compared with the traditional scheme of using cloud computing method to solve medical record sharing problem. The proposed method provides a good application demonstration for solving the security problems in the data sharing process across medical institutions and a reference for cross-domain identity verification in the process of sharing data by using decentralization and auditability of blockchain technology.

Functional implementation of modern safety critical systems is increasingly dependent on software. As a result, software security is very important to system security, and the complexity of software makes it difficult to capture the dangers of component interactions by traditional security analysis methods. In order to ensure the security of safety critical systems, a software security verification method based on System Theoretic Process Analysis (STPA) was proposed. On the basis of the security control structure, by constructing the process model with software process model variables, the system context information of dangerous behavior occurrence was specified and analyzed, and the software security requirements were generated. Then, through the landing gear control system software design, the software security verification was carried out by the model checking technology. The results show that the proposed method can effectively identify the potential dangerous control paths in the software at the system level and reduce the dependence on manual analysis.

For the inefficiency and imbalance between robustness and imperceptibility of most of the current audio watermarking algorithms when applied to MP3 audio, a watermarking algorithm in compressed domain based on low frequency energy of channels of MP3 frames was proposed. The watermarking can be embedded and extracted during MP3 compression and decompression processes, which greatly enhances the efficiency. Considering the good stability of low frequency energy, the low frequency energy of channels was calculated by using Modified Discrete Cosine Transform (MDCT) coefficients produced in MP3 encoding and decoding processes, then the ratio between the energy of left and right channels was quantized with fixed step, and the watermarking was embedded by modifying some MDCT coefficients according to quantified results. Meanwhile, with the proportion of energy in different scalefactor bands, the embedding bands were selected before calculating low frequency energy of channels, which ensured a good balance between robustness and imperceptibility. The experimental results show that the proposed algorithm has a good robustness against various types of attacks while maintaining the original audio quality, especially against MP3 recompression attacks.

Aiming at the problems that traditional aquatic animal image recognition methods have complex steps, low accuracy and poor generalization, and it is difficult to develop Deep Convolutional Neural Network (DCNN) model, a method based on parameter transfer strategy using fine-tune to retrain pre-trained model was proposed. Firstly, the image was preprocessed by data enhancement and so on. Secondly, on the basis of modifying the source model's fully connected classification layer, the weights of high-level convolution modules were set to be trained for adaptive adjustment. Finally, using training time and recognition accuracy on validation set as the evaluation indexes, the performance experiments were conducted on various network structures and different proportion of trainable parameters. The experimental results show that the highest retrained model classification accuracy can reach 97.4%, 20 percentage points higher than the source model, the ideal performance can be obtained when the proportion of trainable parameters is around 75%. It is proved that the fine-tune method can obtain a deep neural network image classification model with good performance under low-cost development condition.

Focusing on the issue of identifying error signatures in Hierarchical Identity Based Cryptography (HIBC) batch verification, a hybrid algorithm of identifying the error signatures was proposed. Firstly, a balanced binary tree was built which used all signatures as the leaves. Secondly, divide-and-conquer and exponent testing methods were used to find error signatures. Meanwhile, the relevance of temporary computing values was used to reduce computing cost. The performance analyses show that the proposed algorithm costs less computation than the individual, the generalized binary splitting, the exponential and the triple pruning search algorithms when there are more than two error signatures. The proposed algorithm can effectively identify error signatures in HIBC batch verification and can be applied in cloud computing authentication.

The original K-means clustering algorithm is seriously affected by initial centroids of clustering and easy to fall into local optima. To solve this problem, an improved K-means clustering algorithm based on Adaptive Cuckoo Search (ACS), namely ACS-K-means, was proposed, in which the search step of cuckoo was adjusted adaptively so as to improve the quality of solution and boost speed of convergence. The performance of ACS-K-means clustering was firstly evaluated on UCI dataset, and the results demonstrated that it surpassed K-means, GA-K-means (K-means based on Genetic Algorithm), CS-K-means (K-means based on Cuckoo Search) and PSO-K-means (K-means based on Particle Swarm Optimization) in clustering quality and convergence rate. Finally, the ACS-K-means clustering algorithm was applied to the development of heat map of urban management cases of Qingxiu district of Nanning city, the results also showed that the proposed method had better quality of clustering and faster speed of convergence.

To improve the recognition accuracy of human activity based on the smart mobile device, an Multiple Classifier Fusion Model for activity recognition (MCFM) based on high reliability weighting was proposed. According to the triaxial acceleration imformation obtained by different smart device with built-in acceleration sensor, those features of high correlation with human daily activities were extracted from the original acceleration as the input of MCFM. Then the three base classifiers of decision tree, Support Vector Machine (SVM) and Back Propagation (BP) neural network were trained for a new fusion classifier by using the High Reliability Weighted Voting (HRWV) algorithm. The experimental results show that the the proposed classifier fusion model can effectively improve the accuracy of human activity recognition, its average recognition accuracy of the five daily activities (stay, walk, run, stairs, downstairs) reaches 94.88%.

The program slicing approach does not describe the suspiciousness of statements, while the coverage analysis based fault localization approach does not analyze the relationship between statements. To solve these problems, a software fault localization approach by statistical analysis of failure context was proposed. Firstly, source code was transformed to an abstract syntax tree and program dependence graphs. Then, instrumentation was performed based on the abstract syntax tree to collect execution information. Next, starting from the failure point, dynamic program slicing based on requirement was conducted in order to get the context of failure. Finally, suspiciousness of nodes in the reverse dynamic program slice was computed, and a dynamic program slice with suspiciousness ranking was output. The proposed approach could not only describe the failure context, but also gave the suspiciousness of the statements. The experimental results show that it has an average 1.3% and 5.6% expense decrease compared with the coverage based analysis approach and the slicing approach respectively, so that it can facilitate the localization and fixing of bugs.

To achieve simple and convenient facial expression recognition, a method combining multi-scale Local Binary Pattern Histogram Fourier (LBP-HF) and Active Shape Model (ASM) was proposed. Firstly, the face regions were detected and segmented by ASM to reduce the influence of unrelated regions, and then LBP-HF were extracted to form recognition vectors. Finally, the nearest neighborhood classifier was applied to recognize expressions. The influences of various scale LBP-HF features on facial expression recognition were studied through extracting LBP-HF features from different scales. At last, multi-scale LBP-HF features were concatenated to discriminate expressions, and more effective expression features were obtained. By comparison with the experimental result of Gabor features, its feasibility and simplication are validated, and the highest mean recognition rate is 93.50%. The experimental results demonstrate that the method can be used for human-computer interaction.

To keep the trade-off of time complexity and accuracy of community detection in complex networks, Community Detection Algorithm based on Clustering Granulation (CGCDA) was proposed in this paper. The granules were regarded as communities so that the granulation for a network was actually the community partition of a network. Firstly, each node in the network was regarded as an original granule, then the granule set was obtained by the initial granulation operation. Secondly, granules in this set which satisfied granulation coefficient were merged by clustering granulation operation. The process was finished until granulation coefficient was not satisfied in the granule set. Finally, overlapping nodes among some granules were regard as isolated points, and they were merged into corresponding granules based on neighbor nodes voting algorithm to realize the community partition of complex network. Newman Fast Algorithm (NFA), Label Propagation Algorithm (LPA), CGCDA were realized on four benchmark datasets. The experimental results show that CGCDA can achieve modularity 7.6% higher than LPA and time 96% less than NFA averagely. CGCDA has lower time complexity and higher modularity. The balance between time complexity and accuracy of community detection is achieved. Compared with NFA and LPA, the whole performance of CGCDA is better.

To solve the sensor node localization problem of Wireless Sensor and Actor Network (WSAN), a range-based localization algorithm with virtual force in WSAN was proposed in this paper, in which mobile actor nodes were used instead of Wireless Sensor Network (WSN) anchors for localization algorithm, and Time Of Arrival (TOA) was combined with virtual force. In this algorithm, the actor nodes were driven under the action of virtual force and made themself move close to the sensor node which sent location request, and node localization was completed by the calculation of the distance between nodes according to the signal transmission time. The simulation results show that the localization success rate of the proposed algorithm can be improved by 20% and the average localization time and cost are less than the traditional TOA algorithm. It can apply to real-time field with small number of actor nodes.

In view of real-time issue of the Hidden Conditional Random Field (HCRF) and marked deviation problem of the Latent-Dynamic Conditional Random Field (LDCRF) during behavior transforming, this article proposed a kind of behavior recognition algorithm based on Stratified Fractal Conditional Random Field (SFCRF). The proposed algorithm improved LDCRF and put forward the concept of score mark, which made the integrity and direction of human behavior specific. The experimental results show that the proposed algorithm can obtain better recognition effect than Conditional Random Field (CRF), HCRF and LDCRF.

In this paper, considering the Web service composition model described by DAG (Directed Acrylic Graph), a new Quality of Service (QoS) computing method for the composition service based on topological sequence reduction (QCMTSR) was proposed. Based on the basic structures and their QoS computing formulas of iterative reduction method two kinds of basic structures (i.e. serial reduction structure and parallel reduction structure) were defined in graph DAG, and their QoS calculation formulas were also given. During accessing each node step by step in the topology sequence for DAG. Repeating this process until the last node in this queue, then the QoS measure results of the last node were the computing results of the composition service. It has been proved that the algorithm can be applied to all the composition services described by DAG, and the experimental results show that the algorithm QCMTSR is more accurate in the measurement of reliability and availability.

Aim at solving the problem of data processing and data demonstration in real-time signals acquisition system in multi-core and multi-thread environment, the authors exploited an efficient algorithm which used bare threads to build data acquisition module and data processing module with a set of optimal threads for allocation in order to make full use of parallel computation to increase the performance and real-time quality of the system. This algorithm worked based on producer-customer and the round-robin scheduling of operating system, referring to the workload of each module to tune the thread allocation solution to make the application achieve better speed-up ratio and real-time quality. The testing result of Field Programmable Gate Array (FPGA) hardware simulation system shows that in dual core environment, this algorithm can produce best combination of threads allocation with which waveform acquisition module and waveform display module can be executed in optimal parallel mode. And compared with other solutions of thread allocation, this solution spends less time in execution. Thus, this algorithm improves the speed-up ratio, computing and the real-time quality of the system. The optimal thread allocation algorithm provides the solution of optimal thread allocation which improves the efficiency of parallel application execution, reduces unnecessary spending of the thread and improves the real-time quality of waveform signal acquisition.

In order to improve the information processing capacity of chaotic neural networks, associative memory performance of a time-delay symmetric global coupled neural network was investigated by using a parameter modulated control method to control the coherent parameter. It can be observed that its output can be stabilized when only partial neurons enter the periodic orbits and the output sequence of the controlled network does not contain other patterns but the stored pattern corresponding to the initial input and its reverse pattern. The experimental results suggest that the network has good tolerance and excellent correct rate so that it is fit for information processing and pattern recognition.

In action classification, methods of feature extraction were either simple with low accuracy, or complicated with poor real-time quality. To resolve this problem, firstly, Enhanced Gait Energy Image (EGEI) was derived from Gait Energy Image (GEI); secondly, high dimensional feature space of the action was reduced to lower dimensional space by Two-Dimensional Locality Preserving Projection (2DLPP); then Nearest-Neighbor (NN) classifier was adopted to distinguish different actions. EGEI could extract more obvious feature information than GEI; 2DLPP outperformed principal component analysis and locality preserving projections in dimensional reduction. It was tested on the Weizmann human action dataset. The experimental results show that the proposed algorithm is simple, achieves higher classification accuracy, and the average recognition ratio reaches 91.22%.

In view of the existing video frequency monitoring technology only relies on the human eye for the examination and lacks intelligence, this article carried out research into the supervisory system based on the gait recognition intelligence. A novel gait representation was proposed. Body silhouette extraction was achieved by background subtraction. A gait cycle was obtained by the correlated signal of the ratio of width and height of human body. Gait energy image was applied on the binary image sequence to construct the feature vector. Then PCA or (2D)2PCA was used to reduce into a low dimension space. The nearestneighbor classifier was adopted to distinguish. This gait recognition method can decrease the influence of the early preprocess effectively; moreover, it makes very good recognition progress under the three visual angles of our gait database in which the camera is put with certain angle of depression.