Virtual Epipolar Line Construction of Single-Lens Bi-Prism Stereovision System

This paper proposes a simple geometrical ray based approach to solve the stereo correspondence problem for the single-lens bi-prism stereovision system. Each image captured using this system can be divided into two sub-images on the left and right and these sub-images are generated by two virtual cameras which are produced by the bi-prism. This stereovision system is equivalent to the conventional two camera system and the two sub-images captured have disparities which can be used to reconstruct back the 3-dimensional （3D） scene. The stereo correspondence problem of this system will be solved geometrically by applying the epipolar geometry constraint on the generated virtual cameras instead of the real CCD camera. Experiments are conducted to validate the proposed method and the results are compared to the calibration based approach to confirm its accuracy and effectiveness.     

An Image Authentication Method by Grouping Most Significant Bits

In 2011,Chan proposed an image authentication method by producing the parity check bits from pixels whose bits have been rearranged.Due to this rearrangement,the value of the most-significant bit of each tampered pixel can be determined according to its parity check bits.With the help of the most-significant bit of the pixel,the pixel can be recovered by selecting two possible(7,4) Hamming code words.However,if the distance between two Hamming code words is within a certain range,incorrect selection may occur.Chan’s method added one additional bit to indicate the correct one.It is trivial that this may degrade the quality of the authenticated image.In this paper,we group four most-significant bits into different groups to form a mapping codebook and the mapping codebook is used to produce authentication data instead of the(7,4) Hamming code book.The experimental results show that the proposed method has a greater ability to recover tampered areas.     

A New Word Clustering Algorithm Based on Word Similarity

Category-based statistic language model is an important method to solve the problem of sparse data in statistical language models. But there are two bottlenecks about this model: 1) The problem of word clustering, it is hard to find a suitable clustering method that has good performance and has not large amount of computation; 2) Class-based method always loses some prediction ability to adapt the text of different domain. In order to solve above problems, a novel definition of word similarity by utilizing mutual information was presented. Based on word similarity, the definition of word set similarity was given and a bottom-up hierarchical clustering algorithm was proposed. Experimental results show that the word clustering algorithm based on word similarity is better than conventional greedy clustering method in speed and performance, the perplexity is reduced from 283 to 207.8.     

Dimensionality Reduction for Hyperspectral Data Based on Sample-Dependent Repulsion Graph Regularized Auto-encoder

To achieve high classification accuracy of hyperspectral data, a dimensionality reduction algorithm called Sample-dependent repulsion graph regularized auto-encoder (SRGAE) is proposed. Based on the sample-dependent graph, by applying the repulsion force to the samples from different classes but nearby, a sampledependent repulsion graph is built to make the samples from the same class will be projected to samples that are close-by and the samples from different classes will be projected to samples that are far away. The sampledependent repulsion graph can avoid the neighborhood parameter selection problem existing in the nearest neighborhood graph. By integrating advantages of deep learning and graph regularization technique, the SRGAE can maintain the learned deep features are consistent with the inherent manifold structure of the original hyperspectral data. Experimental results on two real hyperspectral data show that, when compared with some popular dimensionality reduction algorithms, the proposed SRGAE can yield higher classification accuracy.     

Near-field time-frequency localization method using sparse representation

This paper presents a novel near-field source localization method based on the time-frequency sparse model.Firstly,the method converts the time domain data of array output into time-frequency domain by time-frequency transform;then constructs sparse localization model by utilizing the specially selected time-frequency points,and finally the greedy algorithms are chosen to solve the sparse problem to localize the source.When the coherent sources exist,we propose an additional iterative selection procedure to improve the estimation performance.The proposed method is suitable for uncorrelated and coherent sources,moreover,the improved estimation accuracy and the robustness to low signal to noise ratio(SNR) are achieved.Simulations results verify the efficiency of the proposed algorithm.     

Wavelet Neural Network for Handwritten Manchu Character Unit Recognition

The Manchu character recognition method based on Manchu character unit is an efficient method. In this method, the recognition accuracy rate of Manchu character unit has great influence on the final recognition result. As new approach to solve this problem,a hybrid wavelet neural network scheme has been developed as an assistant method combine with the original combo - distance method. Due to the properties of the wavelet neural network,the training problem can be transformed into a convex optimization process,therefore the global minimum can be obtained and the learning speed is increases. Both the learning samples set and testing samples set are used, experimental results demonstrate the combine method based on the wavelet neural network is more efficient than the single combo - distance method.     

Hybrid System for Robust Faces Detection

The automatic detection of faces is a very important problem. The effectiveness of biometric authentication based on face mainly depends on the method used to locate the face in the image. This paper presents a hybrid system for faces detection in unconstrained cases in which the illumination, pose, occlusion, and size of the face are uncontrolled. To do this, the new method of detection proposed in this paper is based primarily on a technique of automatic learning by using the decision of three neural networks, a technique of energy compaction by using the discrete cosine transform, and a technique of segmentation by the color of human skin. A whole of pictures （faces and no faces） are transformed to vectors of data which will be used for learning the neural networks to separate between the two classes. Discrete cosine transform is used to reduce the dimension of the vectors, to eliminate the redundancies of information, and to store only the useful information in a minimum number of coefficients while the segmentation is used to reduce the space of research in the image. The experimental results have shown that this hybridization of methods will give a very significant improvement of the rate of the recognition, quality of detection, and the time of execution.     

Indoor localization via -graph regularized semi-supervised manifold learning

In this paper,a new l1-graph regularized semi-supervised manifold learning(LRSML) method is proposed for indoor localization.Due to noise corruption and non-linearity of received signal strength(RSS),traditional approaches always fail to deliver accurate positioning results.The l1-graph is constructed by sparse representation of each sample with respect to remaining samples.Noise factor is considered in the construction process of l1-graph,leading to more robustness compared to traditional k-nearest-neighbor graph(KNN-graph).The KNN-graph construction is supervised,while the l1-graph is assumed to be unsupervised without harnessing any data label information and uncovers the underlying sparse relationship of each data.Combining KNN-graph and l1-graph,both labeled and unlabeled information are utilized,so the LRSML method has the potential to convey more discriminative information compared to conventional methods.To overcome the non-linearity of RSS,kernel-based manifold learning method(K-LRSML) is employed through mapping the original signal data to a higher dimension Hilbert space.The efficiency and superiority of LRSML over current state of art methods are verified with extensive experiments on real data.     

A TRUST REGION METHOD FOR MICROWAVE TOMOGRAPHY

A trust region method is proposed to solve the problem of microwave tomography, which is very difficult to be solved for its ill-posedness and nonlinearity. Compared with the Levenberg-Marquardt method, this method introduces more a priori knowledge and might obtain better results, though the two methods are equal in some cases.     

Study on frequency estimation based on weighted least square method with three Fourier coefficients

In this paper, a sinusoidal signal frequency estimation algorithm is proposed by weighted least square method. Based on the idea of Provencher, three biggest Fourier coefficients in the maximum periodogram are considered, the Fourier coefficients can be written as three equations about the amplitude, phase, and frequency, and the frequency is estimated by solving equations. Because of the error of measurement, weighted least square method is used to solve the frequency equation and get the signal frequency. It is shown that the proposed estimator can approach the Cramer-Rao Bound （CRB） with a low Signal-to-Noise Ratio （SNR） threshold and has a higher accuracy.     

Exploring Semantic Association Rules Between Keyword and Visual Feature Clusters for Web Image Retrieval

The semantic gap is a big challenge in image retrieval area. Previous studies in web image retrieval have mainly focused on Relevance feedback （RF） and Latent semantic indexing （LSI） to alleviate the gap. This paper proposes an approach base on Frequent itemset mining （FIM） and Association rule （AR） techniques, which explores the semantic association rule between the two modalities that are represented by keyword and visual feature clusters. The rules are obtained oftline based on the inverted file, and utilized in query process online to realize the integration of the two modalities of web im- ages. Our approach improves the retrieval performance and is scalable well, as well as satisfies the requirement of the web users with no additional interactions. The exper- iments are carried out in our web image retrieval system named VAST （VisuAl ＆ SemanTic image search）, and the results show the effectiveness of the proposed approach.     

Channel Mismatches Correction in Time-Interleaved ADCs Based on Hybrid Filter Bank Reconstruction

Time-interleaved analog-to-digital con- verter （TIADC） is an efficient way to achieve higher sam- pling rate for medium-to-high resolution applications. The performance of a TIADC suffers from the mismatch errors among the sub-channels. This paper presents a method to estimate the channel mismatches using the sub-channels output data. The proposed method introduces an equiv- alent transfer function for each channel to model and es- timate the mismatch errors. A Hybrid filter bank （HFB） structure is used to both model the TIADC and recon- struct the desired uniformly sampled sequence based on the perfect reconstruction conditions of the HFB system. A four-channel 12-bit 400MHz TIADC has been imple- mented in hardware to verify the proposed calibration method. The measured results show that the Spurious- free dynamic range （SFDR） can be improved up to 74dB after being corrected with 64-tap Finite-impulse response （FIR） filters.     

Analysis and Design of Multi-aspect SAR System for Compressive Sensing-Based 3D Imaging

This paper focuses on the analysis and design of Multl-aspect SAR （MuSAR） system for Compressive sensing-based （CS-based） 3D imaging. For this purpose,the Point ambiguous function （PAF） is proposed to analyze the factors that dominate the Mutual coherence （MC） of MuSAR sensing matrix. The PAF contacts with the parameters and configuration of MuSAR system directly and is easy to manipulate. With PAF, the present study analyzes the factors that dominate the performance of CS-based MuSAR 3D imaging. First of all, the stochastic waveform is an excellent selection. Second, the angular-frequency-diversity can improve the robustness of 3D imaging. Finally, the finer sampling of received data could improve the robustness of MuSAR 3D imaging. Simulation experiments show the validity of conclusion.     

A Robust Novel Technique for SPICE Simulation of ESD Snap-back Characteristics

This paper presents a robust and novel technique for the circuit simulation of ESD （ElectroStatic discharge） snap-back characteristics. A new linearization scheme by introducing current as independent variable for the avalanche current model in ESD evaluation shows a good convergence behavior during ESD stress simulation. This technique is compatible with the traditional circuit simulator based on the Modified nodal analysis （MNA） like SPICE. We have implemented the well known Amerasekera＇s ESD MOSFET model in SPICE3fS. The commonly used ESD protection configurations such as GGNMOS （Gate-grounded NMOS） and GCNMOS （Gatecoupled NMOS） are simulated and the simulation results demonstrated the good convergence behavior of this new technique.     

Parameterized Integrated Power and Performance （PIPP） Model for Ultra High-Performance of TOPS level DSP

Amdahl＇s law is a simple and fundamen- tal tool for understanding the evolution of performance as a function of parallelism. Following a recent trend on timing and power analysis of general purpose many-core chip using this law, we develop a novel PIPP analytical model for evaluating the performance and power of hier- archical on-chip large-scale parallel architectures with the core number, super-node size, processing element number, and function unit number taken into consideration. We thereby investigate the influence of workload characteris- tics （Thread-level parallel TLP, Instruction-level parallel ILP and Data-level parallel DLP） on resource allocation with the restriction of performance and power. The re- sults provide some feasible options to design TOPS level DSP architecture as well as a theoretical basis for making the design more scalable.     

Joint Antenna Selection and Robust Beamforming Design in Multi-cell Distributed Antenna System

Most of studies on Distributed Antenna System （DAS） focus on maximizing the sum capacity and perfect channel state information at transmitter （CSIT）. However, CSI is inevitable imperfect in practical wireless networks. Based on the sources of error, there are two models. One assumes error lies in a bounded region, the other assumes random error. Accordingly, we propose two joint antenna selection （AS） and robust- beamforming schemes aiming to minimize the total transmit power at antenna nodes subject to quality of service （QoS） guarantee for all the mobile users （MUs） in multicell DAS. This problem is mathematically intractable. For the bounded error model, we cast it into a semidefinite program （SDP） using semidefinite relaxation （SDR） and S-procedure. For the second, we first design outage constrained robust beamforming and then formulate it as an SDP based on the Bernstein-type inequality, which we generalize it to the multi-cell DAS. Simulation results verify the effectiveness of the proposed methods.     

HGCUDF： Hierarchical Grid Clustering Using Data Field

A new clustering algorithm of Hierarchi- cal grid clustering using data field （HGCUDF） is pro- posed. Under the distributed characteristics of data points on objects, the hierarchical grids divide and conquer the large datasets in their hierarchical subsets, which reduces the scope in search of the clustering centers, and mini- ties the area of data space for generating data field. The compared experiments show that HGCUDF computes the grids rather than retrieves all data from database, for im- proving the efficiency.     

The Price of Anarchy of Network Coding and Routing Based on an ACS Pricing Mechanism

To emphasize the decisions of all users, and the total number of users sharing the same technique, we propose a novel Average cost sharing （ACS） pricing mechanism to study the game between Network coding （NC） and routing flows sharing a single link when users are price anticipating. We characterize the worst-case efficiency bounds of the game compared with the optimal （i.e., the Price of anarchy （PoA））, which can be as low as 4/9 when the number of routing users becomes sufficiently large. NC cannot improve the PoA significantly under ACS. However, to achieve a more efficient use of limited resources, this approach indicates the sharing users have a greater tendency to choose NC. However, the users will follow the majority users＇ choice of data transmission technique.     

An Enhanced IRC Algorithm for LTE Downlink Receiver in Multi-cell Environment

This paper proposes an enhanced Interfer- ence rejection combining （IRC） algorithm for Long term evolution （LTE） downlink receiver in multi-cell communi= cation systems. In this algorithm, a proper Multiple input multiple output （MIMO） receive method is adopted ac- cording to Generalized likelihood ratio test （GLRT） inter- cell interference detection. Iteration between channel es- timation and data detection is carried out to improve the performance of IRC algorithm. Simulation results show that this proposed algorithm can ei~ectively detect inter- cell interference and improve Block error rate （BLER） performance and channel estimation Mean squared error （MSE） compared to non-iterative IRC algorithm, making it suitable for LTE downlink receiver in multi-cell cellular systems.     

Distributed Compressive Video Sensing with Adaptive Measurements Based on Structural Similarity

This paper presents a Distributed compressive video sensing scheme with Adaptive measurements （DCVS-AM）. In this approach, the key frame in each Group of pictures （GOP） is coded by Compressive sensing （CS） with a fixed measurement rate; whereas other frames in the same GOP are compressed by an adaptive random projection in two stages, yielding the Adaptive compressive sensing （ACS） frames. The first stage uses a small and fixed measurement rate and recovers a coarse version. In the second stage, each coarse-version ACS-frame together with its proceeding and following key frames will go through a joint analysis at the decoder side and the analysis result - Structural similarity （SSIM） that is based on a motion-guided interpolation and calculated in a multilevel discrete wavelet transform domain - is sent back to the encoder side to facilitate a re-sampling of the ACS-frame with an adaptive measurement rate. Experimental results show that our proposed DCVS-AM consistently outperforms the state-of-the-art DCVS with a fixed measurement.