For the application of reversible watermarking technology in three-Dimensional (3D) medical images, a 3D medical image reversible watermarking algorithm based on unidirectional prediction error expansion was proposed. Firstly, the intermediate pixels were predicted according to the 3D gradient changes between them and their neighborhood pixels to obtain the prediction errors. Then, considering the features of the 3D medical image generated by magnetic resonance imaging, the external information was embedded into the 3D medical image by combining unidirectional histogram shifting with prediction error expansion. Finally, the pixels were re-predicted to extract the external information and restore the original 3D image. Experimental results on MR-head and MR-chest data show that compared with two-dimensional (2D) gradient-based prediction, the mean absolute deviation of prediction error produced by 3D gradient-based prediction are reduced by 1.09 and 1.40, respectively; and the maximal embedding capacity of each pixel is increased by 0.0456 and 0.1291 bits, respectively. The proposed algorithm can predict the pixels more accurately and embed more external information, so it is applicable to 3D medical image tempering detection and privacy protection for patients.

Among the dictionary-based Super-Resolution (SR) algorithms, the Anchored Neighborhood Regression (ANR) algorithm has been attracted widely attention due to its superior reconstruction speed and quality. However, the anchored neighborhood projections of ANR are unstable to cover varieties of mapping relationships. Aiming at the problem, an image SR algorithm based on adaptive anchored neighborhood regression was proposed, which adaptively calculated the neighborhood center based on the distribution of samples in order to pre-estimate the projection matrix based on more accurate neighborhood. Firstly, K-means clustering algorithm was used to cluster the training samples into different clusters with the image patches as centers. Then, the dictionary atoms were replaced with the cluster centers to calculate the corresponding neighborhoods. Finally, the neighborhoods were applied to pre-compute the projection matrix from LR space to HR space. Experimental results show that the average reconstruction performance of the proposed algorithm on Set14 is better than that of other state-of-the-art dictionary-based algorithms with 31.56 dB of Peak Signal-to-Noise Ratio (PSNR) and 0.8712 of Structural SIMilarity index (SSIM), and even is superior to the Super-Resolution Convolutional Neural Network (SRCNN) algorithm. At the same time, in terms of the subjective performance, the proposed algorithm produces sharp edges in reconstruction results with little artifacts.

Spectral interference in Red Green Blue-Near InfRared (RGB-NIR) images captured by single sensor results in colour distortion and detail information ambiguity of the reconstructed standard Red Green Blue (RBG) and Near InfRared (NIR) images. To resolve this problem, a demosaicing method based on deep learning was proposed. In this method, the grandient dppearance and dispersion problems were solved by introducing long jump connection and dense connection, the network was easier to be trained, and the fitting ability of the network was improved. Firstly, the low-level features such as pixel correlation and channel correlation of the mosaic image were extracted by the shallow feature extraction layer. Secondly, the obtained shallow feature graph was input into successive and multiple residual dense blocks to extract the high-level semantic features aiming at the demosaicing. Thirdly, to make full use of the low-level features and high-level features, the features extracted by multiple residual dense blocks were combined. Finally, the RGB-NIR image was reconstructed by the global long jump connection. Experiments were performed on the deep learning framework Tensorflow using three public data sets, the Common Image and Visual Representation Group (IVRG) dataset, the Outdoor Multi-Spectral Images with Vegetation (OMSIV) dataset, and the Forest dataset. The experimental results show that the proposed method is superior to the RGB-NIR image demosaicing methods based on multi-level adaptive residual interpolation, convolutional neural network and deep residual U-shaped network.

Traditional TSV (Through Silicon Via) table in vertically partially connected three-Dimensional Network-on-Chip (3D NoC) only stores TSV address information, which easily causes network congestion. In order to solve this problem, a record table architecture was proposed. The record table stored not only the nearest four TSV addresses to the router, but also the input-buffer occupancy and fault information of the corresponding router. Based on the record table, a novel adaptive unicast routing algorithm for the shortest transmission path was proposed. Firstly, the coordinates of current node and destination node were calculated to determine the transmission mode of packets. Secondly, by using the proposed algorithm, whether the transmission path was faulty and got information of buffer occupancy was obtained simultaneously. Finally, the optimal transmission port was determined and the packets were transmitted to the neighboring router. The experimental results under two network sizes show that the proposed algorithm has obvious advantages in average delay and throughput compared with Elevator-First algorithm. Additionally, the rates of losing packet under Random model and Shuffle traffic model are 25.5% and 29.5% respectively when the network fault rate is 50%.

To solve the problem existing in traffic classification that Peer-to-Peer (P2P) traffic is much more than that of non-P2P, a new traffic classification method for imbalanced network data was presented. By introducing and improving Synthetic Minority Over-sampling Technique (SMOTE) algorithm, a Mean SMOTE (M-SMOTE) algorithm was proposed to realize the balance of traffic data. On the basis of this, three kinds of machine learning classifiers:Random Forest (RF), Support Vector Machine (SVM), Back Propagation Neural Network (BPNN) were used to identify the various types of traffic. The theoretical analysis and simulation results show that, compared with the imbalanced state, the SMOTE algorithm improves the recognition accuracy of non-P2P traffic by 16.5 percentage points and raises the overall recognition rate of network traffic by 9.5 percentage points. Compared with SMOTE algorithm, the M-SMOTE algorithm further improves the recognition rate of non-P2P traffic and the overall recognition rate of network traffic by 3.2 percentage points and 2.6 percentage points respectively. The experimental results show that the way of imbalanced data classification can effectively solve the problem of low P2P traffic recognition rate caused by excessive P2P traffic, and the M-SMOTE algorithm has higher recognition accuracy rate than SMOTE.

The Graph Cuts segmentation algorithm is not sufficient to segment Printed Circuit Board (PCB) images with non-uniform gray level and complicated inner structure. A new interactive Local Progressive Graph Cuts (LPGC) segmentation method that modeled local constrained energy into a graph cuts framework was presented in this paper. The local constrained energy was adaptively generated by modeling the users additional information behind the interaction, such as the location of the seed, the class of the seed and the relative position between the seed and previous result. Through a comparison of different PCB image segmentation experiments, the results demonstrate the proposed method has better performance compared with the-state-of-art method such as graph cuts in terms of segmentation accuracy, controllability, and user experience.

To deal with the channel uncertainty of the correlated block fading channels, a joint estimation-decoding approach based on Factor Graph Expectation Maximization (FGEM) algorithms was proposed. In the receiver, a message passing method on factor graph was adopted to jointly estimate the channel and decode the message. EM algorithm was used to remove the effect of loops on the convergence of message passing. It also solved the calculation problem of Gaussian mixture message. The calculation of message passing was simplified by the Kalman forward-backward algorithm, which resulted in reduced complexity in joint estimation-decoding. The simulation results show that the proposed algorithm can improve the accuracy of the channel estimation and improve the decoding performance.

To improve the reconstruction speed of the FDK algorithm, this paper presented a fast algorithm based on the graphics processing unit (GPU). The method acquired higher computational efficiency through more careful optimization techniques, including reasonable mode of thread assigning, collecting and pre-computing the variables which were irrelevant with the voxel and the decreasing of number of global memory accesses. The simulation results show that while the fully optimized algorithm makes no precision reduction, the reconstruction time for 2563 is only 0.5 seconds and for 5123 is only 2.5 seconds, which is a big advance in comparison with the latest research findings.

To resist desynchronization attacks, a watermarking scheme using Rational Dither Modulation (RDM) based on Phase-Locked Loop (PLL) was proposed. This scheme took account of two kinds of desynchronization attacks: translation and scaling. Watermark was embedded by RDM and detected by the framework of joint synchronization and decoding. The detector adjusted the estimated attack parameters in PLL according to the results of RDM decoder. Then the new estimated parameters were used to aid watermark decoding, so as to improve the decoding performance. The authors tested the performance of this watermarking system in different data models (DA/DD) and different interpolation algorithms. The simulation results show that this algorithm can resist translation and scaling attacks effectively.