Image classification is an essential task in content-based image retrieval. However, due to the semantic gap between low-level visual features and high-level semantic concepts, and the diversification of Web images, the performance of traditional classification approaches is far from users’ expectations. In an attempt to reduce the semantic gap and satisfy the urgent requirements for dimensionality reduction, high-quality retrieval results, and batch-based processing, we propose a hierarchical image manifold with novel distance measures for calculation. Assuming that the images in an image set describe the same or similar object but have various scenes, we formulate two kinds of manifolds, object manifold and scene manifold, at different levels of semantic granularity. Object manifold is developed for object-level classification using an algorithm named extended locally linear embedding (ELLE) based on intra- and inter-object difference measures. Scene manifold is built for scene-level classification using an algorithm named locally linear submanifold extraction (LLSE) by combining linear perturbation and region growing. Experimental results show that our method is effective in improving the performance of classifying Web images.

Extracting moving targets from video accurately is of great significance in the field of intelligent transport. To some extent, it is related to video segmentation or matting. In this paper, we propose a non-interactive automatic segmentation method for extracting moving targets. First, the motion knowledge in video is detected with orthogonal Gaussian-Hermite moments and the Otsu algorithm, and the knowledge is treated as foreground seeds. Second, the background seeds are generated with distance transformation based on foreground seeds. Third, the foreground and background seeds are treated as extra constraints, and then a mask is generated using graph cuts methods or closed-form solutions. Comparison showed that the closed-form solution based on soft segmentation has a better performance and that the extra constraint has a larger impact on the result than other parameters. Experiments demonstrated that the proposed method can effectively extract moving targets from video in real time.

The automated fare collection (AFC) system, also known as the transit smart card (SC) system, has gained more and more popularity among transit agencies worldwide. Compared with the conventional manual fare collection system, an AFC system has its inherent advantages in low labor cost and high efficiency for fare collection and transaction data archival. Although it is possible to collect highly valuable data from transit SC transactions, substantial efforts and methodologies are needed for extracting such data because most AFC systems are not initially designed for data collection. This is true especially for the Beijing AFC system, where a passenger’s boarding stop (origin) on a flat-rate bus is not recorded on the check-in scan. To extract passengers’ origin data from recorded SC transaction information, a Markov chain based Bayesian decision tree algorithm is developed in this study. Using the time invariance property of the Markov chain, the algorithm is further optimized and simplified to have a linear computational complexity. This algorithm is verified with transit vehicles equipped with global positioning system (GPS) data loggers. Our verification results demonstrated that the proposed algorithm is effective in extracting transit passengers’ origin information from SC transactions with a relatively high accuracy. Such transit origin data are highly valuable for transit system planning and route optimization.

The K-means method is a well-known clustering algorithm with an extensive range of applications, such as biological classification, disease analysis, data mining, and image compression. However, the plain K-means method is not fast when the number of clusters or the number of data points becomes large. A modified K-means algorithm was presented by Fahim et al. (2006). The modified algorithm produced clusters whose mean square error was very similar to that of the plain K-means, but the execution time was shorter. In this study, we try to further increase its speed. There are two rules in our method: a selection rule, used to acquire a good candidate as the initial center to be checked, and an erasure rule, used to delete one or many unqualified centers each time a specified condition is satisfied. Our clustering results are identical to those of Fahim et al. (2006). However, our method further cuts computation time when the number of clusters increases. The mathematical reasoning used in our design is included.

Based on HIT/DLR (Harbin Institute of Technology/Deutsches Zentrum für Luft- und Raumfahrt) Prosthetic Hand II, an anthropomorphic controller is developed to help the amputees use and perceive the prosthetic hands more like people with normal physiological hands. The core of the anthropomorphic controller is a hierarchical control system. It is composed of a top controller and a low level controller. The top controller has been designed both to interpret the amputee’s intensions through electromyography (EMG) signals recognition and to provide the subject–prosthesis interface control with electro-cutaneous sensory feedback (ESF), while the low level controller is responsible for grasp stability. The control strategies include the EMG control strategy, EMG and ESF closed loop control strategy, and voice control strategy. Through EMG signal recognition, 10 types of hand postures are recognized based on support vector machine (SVM). An anthropomorphic closed loop system is constructed to include the customer, sensory feedback system, EMG control system, and the prosthetic hand, so as to help the amputee perform a more successful EMG grasp. Experimental results suggest that the anthropomorphic controller can be used for multi-posture recognition, and that grasp with ESF is a cognitive dual process with visual and sensory feedback. This process while outperforming the visual feedback process provides the concept of grasp force magnitude during manipulation of objects.

The problem of finding stabilizing controllers for switched systems is an area of much research interest as conventional concepts from continuous time and discrete event dynamics do not hold true for these systems. Many solutions have been proposed, most of which are based on finding the existence of a common Lyapunov function (CLF) or a multiple Lyapunov function (MLF) where the key is to formulate the problem into a set of linear matrix inequalities (LMIs). An alternative method for finding the existence of a CLF by solving two sets of linear inequalities (LIs) has previously been presented. This method is seen to be less computationally taxing compared to methods based on solving LMIs. To substantiate this, the computational ability of three numerical computational solvers, LMI solver, cvx, and Yalmip, as well as the symbolic computational program Maple were tested. A specific switched system comprising four second-order subsystems was used as a test case. From the obtained solutions, the validity of the controllers and the corresponding CLF was verified. It was found that all tested solvers were able to correctly solve the LIs. The issue of rounding-off error in numerical computation based software is discussed in detail. The test revealed that the guarantee of stability became uncertain when the rounding off was at a different decimal precision. The use of different external solvers led to the same conclusion in terms of the stability of switched systems. As a result, a shift from using a conventional numerical computation based program to using computer algebra is suggested.

We propose a feedback type oscillator and two negative resistance oscillators. These microwave oscillators have been designed in the S band frequency. A relatively symmetric resonator is used in the feedback type oscillator. The first negative resistance oscillator uses a simple lumped element resonator which is substituted by a microstrip resonator in the second oscillator to improve results. The negative resistance oscillator produces 4.207 dBm and 7.124 dBm output power with the lumped element resonator and microstrip resonator respectively, and the feedback type oscillator produces ?10.707 dBm output power. The feedback type oscillator operates at 3 GHz with phase noise levels at ?83.30 dBc/Hz and ?103.3 dBc/Hz at 100 kHz and 1 MHz offset frequencies respectively. The phase noise levels of the negative resistance oscillator with the lumped element resonator are ?94.64 dBc/Hz and ?116 dBc/Hz at 100 kHz and 1 MHz offset frequencies respectively, at an oscillation frequency of 3.053 GHz. With the microstrip resonator the phase noise levels are ?99.49 dBc/Hz and ?119.641 dBc/Hz at 100 kHz and 1 MHz offset frequencies respectively, at an oscillation frequency of 3.072 GHz. The results showed that both the output power and the phase noise of the negative resistance oscillators were better than those of the feedback type oscillator.