Haitao Zhao

Rapid and brief communication: Local structure based supervised feature extraction

In the past few years, the computer vision and pattern recognition community has witnessed the rapid growth of a new kind of feature extraction method, the manifold learning methods, which attempt to project the original data into a lower dimensional feature space by preserving the local neighborhood structure. Among them, locality preserving projection (LPP) is one of the most promising feature extraction techniques. However, when LPP is applied to the classification tasks, it shows some limitations, such as the ignorance of the label information. In this paper, we propose a novel feature extraction method, called locally discriminating projection (LDP). LDP utilizes class information to guide the procedure of feature extraction. In LDP, the local structure of the original data is constructed according to a certain kind of similarity between data points, which takes special consideration of both the local information and the class information. The similarity has several good properties which help to discover the true intrinsic structure of the data, and make LDP a robust technique for the classification tasks. We compare the proposed LDP approach with LPP, as well as other feature extraction methods, such as PCA and LDA, on the public available data sets, FERET and AR. Experimental results suggest that LDP provides a better representation of the class information and achieves much higher recognition accuracies.

Fiber Bragg grating-based plane strain monitoring of aerostat envelope structures

A theoretical analysis of fiber Bragg grating (FBG)-based plane strain monitoring of aerostat envelope structures is presented. Plane strain analysis of FBG-based aerostat envelope structures is much more complex than the case along the axis of the optical fiber because the effect of transverse stress on the FBG should be taken into consideration. To achieve accurate strain measurement of the aerostat envelope, a theoretical model is set up by using two perpendicular fibers in the monitoring. An analytical formula that evaluates the relationship between the strain measured by FBG sensors and the real one in the aerostat envelope is established. On the other hand, the real strain of aerostat envelope strain is affected by two unknown parameters, axial transfer rate K(L) and the radial transfer rate K(R). An equation is derived to calculate the axial transfer rate K(L). Then, the finite element method results show that K(R) is a very small value, but it cannot be ignored in accurate measurement. This paper would lay a theoretical groundwork for the research and design of FBG sensors in the structural health monitoring of aerostat envelope structures.

New method for dynamic bias estimation: Gaussian mean shift registration

A novel algorithm, Gaussian mean shift registration (GMSR), is proposed for multisensor dynamic bias estimation. The sufficient condition for convergence of a Gaussian mean shift procedure is given, which extends the current theorem from a strictly convex kernel to a piece-wise convex and concave kernel. The Gaussian mean shift algorithm combined with the extended Kalman filter (EKF) is implemented to estimate the dynamic bias based on the measurements from a single target, which is an iterative optimization procedure. Monte Carlo simulations show that the new algorithm has significant improvement in performance with reducing root mean square (RMS) errors compared with the minimum mean square error (MMSE) estimator, based on multiple targets and multiple frames. The proposed estimator is close to the theoretical lower bound, i.e., it is more efficient in estimating the dynamic bias than other methods.

Local-information-based uncorrelated feature extraction

In the past few years, the computer vision and pattern recognition community has witnessed the rapid growth of a new kind of feature extraction method, the manifold learning methods, which attempt to project the original data into a lower dimensional feature space by preserving the local neighborhood structure. Among them, locality preserving projection (LPP) is one of the most promising feature extraction techniques. Based on LPP, we propose a novel feature extraction method, called uncorrelated locality preserving projection (ULPP). We show that the extracted features via ULPP are statistically uncorrelated, which is desirable for many pattern analysis applications. We compare the proposed ULPP approach with LPP and principal component analysis (PCA) on the publicly available data sets, FERET and AR. Experimental results suggest that the proposed ULPP approach provides a better representation of the data and achieves much higher recognition accuracies.

Configuration Analysis of a High-Altitude Airship’s Regenerative Power System

AbstractIn the design of the power system of a high-altitude airship (HAA), the principal target is to satisfy the needs of the power demand and achieve the best balance of minimum weight and maximum reliability. To handle this problem, configuration analysis of the power system is performed. Mathematical models of output power, reliability, and weight are presented. Relationships between weight, reliability, and configurations are discussed in detail. Several design rules related to the design of the HAA’s power system are deduced. For obtaining the optimal configuration, the self-adaptive genetic algorithm (GA) is applied. Results show that the optimal configuration, compared to the configuration without redundancy, has a 258% increase in reliability and a 55.9% increase in weight. The weight increase is necessary to achieve more reliability improvement.

Force identification based on a comprehensive approach combining Taylor formula and acceleration transmissibility

Abstract Force identification is a crucial inverse problem in structural dynamics. In this paper, a new method integrating Taylor formula algorithm and acceleration transmissibility concept is put forward to identify the magnitude and location of loads in time domain. The Taylor formula algorithm expresses the response vectors as Taylor-series expansion and then, a series of deductions are implemented. Ultimately an explicit discrete equation which associates output acceleration response, structure characteristic and input excitation together is established. After establishing the explicit discrete equation, acceleration transmissibility concept is utilized to identify the location of forces. Under the premise of knowing the acceleration response and structure characteristic, force magnitude can be calculated. To verify the effectiveness of proposed method, one builds up a theoretical simulation model in which different types of dynamic excitations are exerted on an inflatable cantilever beam. Meanwhile, classical state space algorithm is made a contrast with Taylor formula algorithm in the step of force magnitude identification. Calculation results demonstrate that the integration method is capable of identifying the location of excitations precisely. Reconstruction of force time history reaches a higher accuracy compared to state space algorithm as well. In addition, the anti-noise ability of proposed algorithm is discussed.

Preliminary reliability analysis of a high-altitude airship’s envelope

To analyze the reliability of an airship’s envelope, the methodology of structural reliability analysis is adopted. The basic theory and the detailed steps of the algorithm of the first-order reliability method are discussed. For finding multiple design points, the method of adding bulge to the limit-state function is applied. With regard to the problem of envelope’s reliability, the safety criterion and limit state function of the airship’s envelope are analyzed. The mathematical model of the envelope’s maximum stress is also presented. The reliability simulation of a stratospheric airship’s envelope is taken as an example. Results of sensitivity analyses of the envelope are also obtained.

Visible-information-aided eyeglasses removing for thermal image reconstruction

Recently, a number of studies have demonstrated that thermal infrared (IR) imagery offers a promising alternative to visible imagery in face recognition problems due to its invariance to visible illumination changes. However, thermal IR has other limitations including that it is opaque to glass. As a result, thermal IR imagery is very sensitive to facial occlusion caused by eyeglasses. Fusion of the visible and thermal IR images is an effective way to solve this problem. In this paper, using the face reconstruction information of the visible images, we propose a nonlinear eyeglasses removing algorithm which can successfully reconstruct the thermal images. Experiments on publicly available data set show the excellent performance of our algorithm.