Image inpainting is a common method of image tampering. Image inpainting methods based on deep learning can generate more complex structures and even new objects， making image inpainting forensics more challenging. Therefore， an end-to-end U-shaped Feature Pyramid Network （FPN） was proposed for image inpainting forensics. Firstly， multi-scale feature extraction was performed through the from-top-to-down VGG16 module， and then the from-bottom-to-up feature pyramid architecture was used to carry out up-sampling of the fused feature maps， and a U-shaped structure was formed by the overall process. Next， the global and local attention mechanisms were combined to highlight the inpainting traces. Finally， the fusion loss function was used to improve the prediction rate of the repaired area. Experimental results show that the proposed method achieves an average F1-score and Intersection over Union （IoU） value of 0.791 9 and 0.747 2 respectively on various deep inpainting datasets. Compared with the existing Localization of Diffusion-based Inpainting （LDI）， Patch-based Convolutional Neural Network （Patch-CNN） and High-Pass Fully Convolutional Network （HP-FCN） methods， the proposed method has better generalization ability， and also has stronger robustness to JPEG compression.

In order to solve the model drift problem of Multi-Domain convolutional neural Network (MDNet) when the target moves rapidly and the appearance changes drastically, a Multi-Domain convolutional neural Network based on Self-Attention (SAMDNet) was proposed to improve the performance of the tracking network from the dimensions of channel and space by introducing the self-attention mechanism. First, the spatial attention module was used to selectively aggregate the weighted sum of features at all positions to all positions in the feature map, so that the similar features were related to each other. Then, the channel attention module was used to selectively emphasize the importance of interconnected channels by aggregating all feature maps. Finally, the final feature map was obtained by fusion. In addition, in order to solve the problem of inaccurate classification of the network model caused by the existence of many similar sequences with different attributes in training data of MDNet algorithm, a composite loss function was constructed. The composite loss function was composed of a classification loss function and an instance discriminant loss function. First of all, the classification loss function was used to calculate the classification loss value. Second, the instance discriminant loss function was used to increase the weight of the target in the current video sequence and suppress its weight in other sequences. Lastly, the two losses were fused as the final loss of the model. The experiments were conducted on two widely used testing benchmark datasets OTB50 and OTB2015. Experimental results show that the proposed algorithm improves success rate index by 1.6 percentage points and 1.4 percentage points respectively compared with the champion algorithm MDNet of the 2015 Visual-Object-Tracking challenge (VOT2015). The results also show that the precision rate and success rate of the proposed algorithm exceed those of the Continuous Convolution Operators for Visual Tracking (CCOT) algorithm, and the precision rate index of it on OTB50 is also superior to the Efficient Convolution Operators (ECO) algorithm, which verifies the effectiveness of the proposed algorithm.

When the query condition involves multiple indexed attributes， TiDB cannot effectively use multiple indexes to generate a more efficient execution plan. After studying the existing solutions of databases， such as PostgreSQL and MySQL， a new type of data access path using multiple indexes simultaneously was proposed in TiDB to solve the problem， namely MultiIndexPath. Firstly， a possible access path named MultiIndexPath for a certain query was generated and its physical plan named MultiIIndexPlan was created， then the cost of this plan was calculated. Secondly， the general execution framework of MultiIndexPath was proposed after combining the architecture and implementation of TiDB. Finally， the Pipeline execution plan was proposed with the condition of conjunctive normal form. The whole work was implemented based on TiDB 3.0 and several experiments were conducted. Experimental results show that the performance of the proposed scheme is improved by at least one order of magnitude compared with the original TiDB when the condition is the disjunctive normal form. With the condition of conjunctive normal form， the performance of the scheme is also better than that of the original TiDB.

China's air pollution control policy has gradually shifted from pollution control to risk prevention and control, and existing air quality monitoring equipment and platform services are limited to environmental monitoring rather than exposure monitoring. Aiming at this problem, a comprehensive visual analysis and decision support platform based on B/S architecture-Air Pollution Exposure Risk Measurement System (APERMS) was designed and developed. Firstly, based on air pollution concentration monitoring data and exposure spatio-temporal behavior patterns, the complete air pollution exposure risk measurement technology route of pollution concentration mapping, individual exposure measurement, population exposure measurement and exposure risk assessment was researched and integrated. Secondly, based on the principle of high availability and reliability, the overall system architecture design, database design and functional modules design were carried out. Finally, GIS and J2EE Web technologies were utilized to complete the development of APERMS, realizing the high spatio-temporal resolution simulation of air pollution concentration distribution, accurate assessment of individual and population exposure of air pollution and comprehensive evaluation of air pollution exposure risk levels. The APERMS is mainly used in the air pollution monitoring and environmental health management industries, to provide effective technical support for risk aversion as well as pollution prevention and control.

Focusing on the problem that the randomness of task assignment is too high and the utility value is not ideal under the spatial crowdsourcing environment, an adaptive threshold algorithm based on statistical prediction was proposed. Firstly, the numbers of free tasks, free workers and free positions in the crowdsourcing platform in real-time was counted to set the threshold value. Secondly, according to the historical statistical analysis, the distributions of tasks and workers were divided into two balanced parts, then the Min-max normalization method was applied to match each task to a certain worker. Finally, the probability of the appearance of the matched workers was calculated to verify the effectiveness of the task distribution. The experimental results on real data show that, compared with random threshold algorithm and greedy algorithm, the utility value of the proposed algorithm was increased by 7% and 10%, respectively. Experimental result indicates that the proposed adaptive threshold algorithm can reduce the randomness and improve the utility value in the process of task assignment.

To reduce the impact of Programmed I/O (PIO) write latency in PCI express (PCIe) transmission process, too many times of interaction between the host and the embedded processing system and other issues on transmission bandwidth, a Direct Memory Access (DMA) controller based on command buffering mechanism was designed to improve the transmission bandwidth utilization. Using the internal command buffer of the Field-Programmable Gate Array (FPGA), the DMA controller could cache the data transfer request of the PC. The FPGA could dynamically access the storage space of the PC according to its own requirements and enhance the transmission flexibility. At the same time, a dynamic mosaic DMA scheduling method was proposed to reduce the times of host-to-hardware interaction and interrupt generation by merging the access requests of adjacent storage areas. In the system transmission rate test, the maximum write speed of DMA was 1631 MB/s, the maximum rate of DMA read was up to 1582 MB/s, the maximum of bandwidth was up to 85.4% of the theoretical bandwidth of PCIe bus. Compared with the traditional PIO mode DMA transfer method, DMA read bandwidth increased by 58%, DMA write bandwidth increased by 36%. The experimental results show that the proposed design can effectively improve the DMA transfer efficiency, and is significantly better than PIO method.

The hardware cost of fast computing of structural topology optimization based on traditional parallel computing method is high and the efficiency of coding development is low. In order to solve the problems, an entire parallel computing method of Bi-directional Evolutionary Structural Optimization (BESO) based on Matlab and Graphics Processor Unit (GPU) was proposed. Firstly, the advantages, disadvantages and application range of three kinds of GPU computing methods used for GPU parallel computing in Matlab were discussed. Secondly, the built-in function was introduced to directly realize parallel computing of the math operations between vector and dense matrix, the MEX function was introduced to realize the fast computing of sparse finite element equations by CUSOLVER library, and the Parallel Thread eXecution (PTX) code was introduced to realize the parallel computing of the optimization decisions of the element sensitivity analysis and other decisions in the topology optimization. The numerical examples show that the GPU parallel computing program based on Matlab has a high coding efficiency, and can avoid the precision difference between the different programming languages. The GPU parallel computing program can ultimately achieve a considerable speedup with the same results.

In order to obtain the sequences with a few autocorrelation values and large linear complexity, a new class of binary cyclotomic sequences of order 3 with period p were constructed, where p is a prime and p≡1(mod 3). The autocorrelation was computed based on cyclotomy, and the condition for p that assures the 3-valued autocorrelation was discussed. The condition is that p should be the form p= a ²+12 for an integer a. The linear complexity is p-1 if p is the form, or 2( p-1)/3 otherwise. By computer experiments, all ps' satisfying the form were found, the corresponding sequences were given, and the autocorrelation and linear complexity were confirmed. The linear complexity was the same as that of the known ternary cyclotomic sequence of order 3. Compared with the related known binary cyclotomic sequences of even order, the linear complexity was the same or better in most cases. The method in this paper can be extended to construct other cyclotomic sequences of odd order with a few autocorrelation values and large linear complexity. Since the cyclotomic sequences of larger odd order also have better balance, they can be applied to stream ciphers and communication systems.

Splicing is the most universal image tampering operation, detection of which is effective for identifying image tamper. A blind splicing detection method was proposed. The method firstly analyzed the effects of different sub-bands on image splicing detection according to features of wavelet transform. High frequency sub-band was verified to be more appropriate for splicing detection both from theory analysis and experiment results. Secondly, the method conducted difference operation, rounded and made threshold to the coefficients as discrete Markov states, and calculated the state transition probabilities as splicing features. Finally, Support Vector Machine (SVM) was used as classifier, and the features were tested on Columbia image splicing detection evaluation datasets. The experimental results show that the proposed method performs better compared with other features and achieves a detection accuracy rate of 94.6% on the color dataset specially.

A new information disguising method based on empirical mode decomposition was proposed in this paper. The pseudorandom sequence controlled by seeds of the SHA-256 one-way hash function was generated, and Digital Elevation Model （DEM） data for disguising was achieved by decomposing the expanded pseudorandom sequence via EMD. The high vision fraudulence was obtained for disguised DEM data. Furthermore, the concept of the generalized histogram for DEM data was also proposed and the watermarking was reversibly embedded in the disguised DEM data by modifying its generalized histogram. The disguised DEM data could be completely reconstructed without any distortion from the marked data after the watermark had been extracted. The secret DEM data could be recovered via the seed. The proposed algorithm is of high security.

This paper first showed respective characteristics of typical database application systems and introduced the theory of sniffer technology.It put forward a new method of analyzing data transfer security numerically with sniffer technology.Then it told how to make an experiment with this method and showed results of the experiment by using this method. At last,the paper emphasized the advantage of using this method to analyze data transfer security of database application system.