Based on the vectorised and cache optimised kernel, a parallel lower upper decomposition with a novel communication avoiding pivoting scheme is developed to solve dense complex matrix equations generated by the method of moments. The fine-grain data rearrangement and assembler instructions are adopted to reduce memory accessing times and improve CPU cache utilisation, which also facilitate vectorisation of the code. Through grouping processes in a binary tree, a parallel pivoting scheme is designed to optimise the communication pattern and thus reduces the solving time of the proposed solver. Two large electromagnetic radiation problems are solved on two supercomputers, respectively, and the numerical results demonstrate that the proposed method outperforms those in open source and commercial libraries.

An algorithm for the de-noising of S-parameter data used in microwave imaging is proposed. The complex S-parameter frequency-sweep data are collected through scans over an acquisition surface and the algorithm separates efficiently the resulting two-dimensional responses (one frequency at a time) into a signal and a noise component. The separation is performed with an iterative procedure similar to the empirical mode decomposition. The signal component estimates the noise-free data, whereas the remaining data content estimates the noise and uncertainty in the measurement. The algorithm performance is verified with measured data.

This study presents a novel human action recognition method based on the sequences of depth maps, which provide additional body shape and motion information for action recognition. First, the authors divide each depth sequence into a number of sub-sequences. All these sub-sequences are of uniform length. By controlling vague boundary (VB), they construct a VB-sequence which consists of an original sub-sequence and its adjacent sequences. Then, each depth frame in a VB-sequence is projected onto three orthogonal Cartesian planes, and the absolute value of the difference between two consecutive projected maps is accumulated to form a depth motion map (DMM) to describe the dynamic feature of a VB-sequence. Finally, they concatenate the DMMs of all the VB-sequences in one video sequence to describe an action. Collectively, they call them VB division of depth model. For classification, they apply robust probabilistic collaborative representation classification. The recognition results applied to the MSR Action Pairs, MSR Gesture 3D, MSR Action3D, and UTD-MHAD datasets indicate superior performance of their method over most existing methods.

Through silicon via (TSV) interconnect reliability is a problem in electronic packaging. The authors address the insertion losses, deflections which can result to separation of TSV layers and hoop stresses. These problems are due to different coefficient of thermal expansion between materials. The authors propose a robust methodology for (TSV) liners in this paper which in turn solves the reliability problem in (TSV). Silicon dioxide material is used in their paper as a TSV liner. First, they modelled the equivalent TSV circuit in advanced design systems (ADS). The authors then simulated it to obtain the TSV characterisation from which they obtained the S-parameter S21 which represents the insertion losses. Insertion losses have been described with changes in frequencies from 0 to 20 GHz with changes in TSV thickness from 7 to 8 µm. Later two different shapes of the TSV liner; the disc- and rod-shaped are modelled in analysis system 14 software. The two shapes with a radius of 5 µm each and a fixed pressure of 100 µPa developed changes in hoop stresses and deflections when the liners thicknesses are varied from 2 to 3 µm. The disc shape experienced least reliability problems so the authors propose its use in via structures.

In recent decades, the local pattern descriptor has achieved tremendous success in the field of face recognition, pedestrian detection, and image texture analysis. This study presents a generic approach, called the filtered local pattern descriptor (FLPD), which expands the traditional local pattern descriptor (TLPD) by using multi-scale and multi-type filter banks. The FLPD encodes the local information of an image based on the convolutional sum of the sub-image blocks and the filter banks, instead of the original pixel values in the TLPD. This design can effectively increase the diversity of the TLPD feature extraction, thereby enhancing the ability of feature representation and its reliability. Two FLPD-based feature representation methods are proposed for the face image and the pedestrian image. To evaluate the performance of the proposed FLPD, extensive experiments on face recognition and infrared pedestrian detection are conducted using several benchmark image datasets. The experimental results illustrate that the FLPD has a significant advantage in the discrimination and stability of feature extraction, and is able to achieve a satisfactory accuracy in comparison with state-of-the-art methods. It is demonstrated that the FLPD is a powerful and convenient extension of the TLPD by filter banks, and suitable to be implemented as feature extraction into approaches to solve the binary or multi-class image classification problems.

In this study, the authors propose a state space modelling approach for trust evaluation in wireless sensor networks. In their state space trust model (SSTM), each sensor node is associated with a trust metric, which measures to what extent the data transmitted from this node would better be trusted by the server node. Given the SSTM, they translate the trust evaluation problem to be a non-linear state filtering problem. To estimate the state based on the SSTM, a component-wise iterative state inference procedure is proposed to work in tandem with the particle filter (PF), and thus the resulting algorithm is termed as iterative PF (IPF). The computational complexity of the IPF algorithm is theoretically linearly related with the dimension of the state. This property is desirable especially for high-dimensional trust evaluation and state filtering problems. The performance of the proposed algorithm is evaluated by both simulations and real data analysis.

This study presents an efficient disease classification approach based on medical images. The approach is more efficient as it reduces the computational complexity. The implementation uses only two wavelet filters in selecting the texture features as compared with five filters used in the earlier research works. The computed average and energy features are fed to feed-forward neural network (FFNN) and support vector machine (SVM) classifiers. The SVM is proved as a better classifier than the FFNN for all the three diseases related to skin, breast and retina with an improved accuracies of 89%, 92% and 100%, respectively. Also, a graphical user interface is developed useful for various disease classification based on the whole dataset of size 100.

For positioning a moving ultrasonic transmitter, most of the existing ultrasonic positioning systems require the use of a bank of correlators to estimate the Doppler shift associated with its movement which require high computational complexity. In this paper, for positioning a moving transmitter, a computationally efficient a Doppler shift estimation and compensation technique is proposed. As the proposed approach has the ability to measure the Doppler shift directly from the received signal, it does not require to use a bank of correlators to estimate the Doppler shift associated with its movement of the transmitter.

The segmentation of the right ventricle in cardiac magnetic resonance imaging is a difficult task due to the blurred border and variable shape. To overcome these problems, an approach using graph cuts (GC) segmentation based on multi-dimensional local features is proposed. The features are reduced by a principal component analysis, which are incorporated into the t-link and the n-link. Experiments are performed on 240 MR images from 16 patients. The segmentation results of the proposed algorithm are obtained using Gabor and histogram of oriented gradient (HOG) features, respectively. For Dice metric, the median value of using Gabor and HOG increases significantly compared to original GC from 0.8012 to 0.8174 and 0.8306 at end-diastole (ED), from 0.6555 to 0.7021 and 0.7124 at end-systole (ES). For Hausdorff distance, the median value of using Gabor and HOG decreases significantly compared to original GC from 12.88 to 12.33 and 11.82 mm at ED, from 17.20 to 12.75 and 13.41 mm at ES. The proposed method outperforms original GC depending on image intensity only. The segmentation errors of using HOG are slightly less than those of using Gabor.

Function generators are vital tools for many industrial applications. Sinusoidal oscillators are the essential units of function generators and investigating their performance is therefore necessary. The Barkhausen criterion is widely applied in designing sinusoidal oscillators because of its simplicity. The authors exploit a crucial necessary and sufficient oscillation criterion for Wien Bridge and Colpitts oscillators based on Simple Chen's Electric Unifying Approach. Their approach resolves the shortcoming of the traditional Barkhausen criterion, namely that it entails only a necessary condition. They propose a simple oscillation determining criterion that uses a stable sine-wave oscillator with an adjustable oscillation frequency. The derived oscillation criterion establishes a precise condition for the existence of oscillatory behaviour and does not require the fixed value used in the traditional Barkhausen criterion. To demonstrate its applicability, the proposed approach for the Wien Bridge and Colpitts oscillators is applied to implement a function generator.