高维空间下求解MDA的扰动算法及其在人脸识别中的应用

本文提出了一种在高维空间下直接求MDA(Multiple discriminant analysis)最佳解的扰动算法，并把这一算法应用于人脸的识别中。在传统的MDA求解算法中，一般要求训练样本的个数足够大，以至于其类内散射矩阵为非奇异矩阵，即所谓的“小样本”问题。但是，在入脸识别中，由于人脸空间的维数非常大，而训练样本的个数有限，造成类内散射矩阵为奇异矩阵，从而使得用传统的求解方法失效。为了能在高维空间中求出MDA的最佳解，本文采用扰动的思想，巧妙地避开了矩阵的奇异性问题并找到了最佳变换矩阵。此外，把这一算法用于人脸的识别中，对ORL人脸图像库的实验显示，采用本文提出的算法达到比较低的错误率，其错误率仅为特征脸(Eigenface)方法的49．9％，为Fisher脸(Fisherface)的7914％。     

人脸识别研究进展

人脸识别技术在各种应用的推动下发展很快。而近几年中，为避免各种实际应用环境的限制，非约束环境下的人脸识别也得到了很大的发展。本文概述了人脸识别的一些主要方法，讨论了影响人脸识别的几个主要因素及其处理方法以及基于视频的人脸识别技术，最后指出了鲁棒的人脸识别技术的研究方向和发展前景。     

Threshold and growth mechanism of fatigue cracks under mode II and III loadings

ABSTRACT The fatigue crack growth behaviour of 0.47% carbon steel was studied under mode II and III loadings. Mode II fatigue crack growth tests were carried out using specially designed double cantilever (DC) type specimens in order to measure the mode II threshold stress intensity factor range, ΔK_IIth. The relationship ΔK_IIth > ΔK_Ith caused crack branching from mode II to I after a crack reached the mode II threshold. Torsion fatigue tests on circumferentially cracked specimens were carried out to study the mechanisms of both mode III crack growth and of the formation of the factory‐roof crack surface morphology. A change in microstructure occurred at a crack tip during crack growth in both mode II and mode III shear cracks. It is presumed that the crack growth mechanisms in mode II and in mode III are essentially the same. Detailed fractographic investigation showed that factory‐roofs were formed by crack branching into mode I. Crack branching started from small semi‐elliptical cracks nucleated by shear at the tip of the original circumferential crack.     

Effects of stimulus emotionality and sentence generation on memory for words in adults with unilateral brain damage.

Does sentence generation and/or stimulus emotionality enhance verbal memory in patients with neurological impairment? This question was addressed by testing 40 patients with unilateral stroke (20 with left-brain and 20 with right-brain damage) and 20 healthy control participants for recall and recognition of 48 target words. During encoding, emotional and nonemotional words were either presented in sentences (read condition) or used to form sentences (generate condition). Both word emotionality and generative processing improved memory performance in all groups. The authors suggest that a similar influence (i.e., cognitive activation) underlies both of these memory-enhancing effects, although the putative origins of the 2 effects are quite different. Neuropsychological underpinnings and clinical implications of these phenomena are discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Damage mechanism in weldment of 2.25Cr-1Mo steel under creep-fatigue loading

The purpose of this study is to clarify damage process of “Type IV cracking” in weldment of a 2.25Cr-1Mo steel and to propose a micro damage prediction method. From continuous observation under a creep-fatigue test, it was found that spherical shape voids initiated and grew on grain boundaries in fine grain region and these voids grow continuously by changing their shape to crack-like. Both spherical and crack-like void growth rate equations were derived from the proposed void growth model. It was indicated that measured void growth rate under the creep-fatigue loading was well predicted by the growth rate equations.     

Aspects of rapid crack propagation in silicon

Fracture experiments with silicon specimens in recent years have shown the need for a new approach to the analysis of rapidly propagating cracks in single crystals. Behaviour and phenomena have been revealed that fracture in these materials is rather different from the fracture of both amorphous and polycrystalline materials. We show that continuum mechanics is insufficient for analyzing crack propagation in single crystals since it is unable to consider atomistic‐scale phenomena. Accordingly, we describe basic phenomena associated with rapid crack propagation in silicon : (i) anisotropic velocity‐dependent R‐curve behaviour, as a key phenomenon dictating atomistic scale behaviour, (ii) crack deflection from one cleavage plane to another as a mesoscopic scale phenomenon in single‐crystal fracture, (iii) the Rayleigh surface wave speed as the limiting crack tip velocity is re‐examined, (vi) the lowest crack velocity in brittle crystals is examined, and finally (v) the interaction between crack path and preferred cleavage planes in single crystals is depicted.     

Elastoplastic modelling of subsurface crack growth in rail/wheel contact problems

Propagation of small subsurface cracks subjected to shear under repeated rolling contact load is studied. An analytical crack model (Dugdale) with plastic strips at the two crack tips is employed. Compressive stresses promoting crack closure and friction between crack faces are considered. The triaxial stress state is used in the yield criterion. A damage criterion is suggested based on experimental LCF data. In a numerical study, critical crack lengths are found below which propagation of an existing crack should be effectively suppressed.     

Off-axis fatigue cracking behaviour in notched fibre metal laminates

The off‐axis fatigue cracking behaviour of notched fibre metal laminates under constant amplitude loading conditions was investigated experimentally and numerically. It was found that the off‐axis fatigue crack initiation life decreased as the off‐axis angles increased. This indicated that the off‐axis laminates raised the applied stress level in the aluminium (Al) layer and subsequently resulted in earlier cracking in the Al layer. The off‐axis fatigue crack initiation lives of notched fibre metal laminates were predicted using lamination theory and an energy‐based critical plane fatigue damage analysis from the literature. After a crack initiated in the Al layer, it was observed that the crack path angles of the off‐axis specimens were neither perpendicular to the fibre nor to the loading direction. A finite‐element model was established for predicting the crack path angles.     

Crack closure in fibre metal laminates

GLARE is a fibre metal laminate (FML) built up of alternating layers of S2-glass/FM94 prepreg and aluminium 2024-T3. The excellent fatigue behaviour of GLARE can be described with a recently published analytical prediction model. This model is based on linear elastic fracture mechanics and the assumption that a similar stress state in the aluminium layers of GLARE and monolithic aluminium result in the same crack growth behaviour. It therefore describes the crack growth with an effective stress intensity factor (SIF) range at the crack tip in the aluminium layers, including the effect of internal residual stress as result of curing and the stiffness differences between the individual layers. In that model, an empirical relation is used to calculate the effective SIF range, which had been determined without sufficiently investigating the effect of crack closure. This paper presents the research performed on crack closure in GLARE. It is assumed that crack closure in FMLs is determined by the actual stress cycles in the metal layers and that it can be described with the available relations for monolithic aluminium published in the literature. Fatigue crack growth experiments have been performed on GLARE specimens in which crack growth rates and crack opening stresses have been recorded. The prediction model incorporating the crack closure relation for aluminium 2024-T3 obtained from the literature has been validated with the test results. It is concluded that crack growth in GLARE can be correlated with the effective SIF range at the crack tip in the aluminium layers, if it is determined with the crack closure relation for aluminium 2024-T3 based on actual stresses in the aluminium layers.     

Real-time control of video surveillance systems with program supervision techniques

In this article, we describe a knowledge-based controlled platform using program supervision techniques. This platform eases the creation and the configuration of video surveillance systems. Several issues need to be addressed to provide a correct system configuration: (1) to choose, among a library of programs, those which are best satisfying a given user request, (2) to assign a correct value for each program parameter, (3) to evaluate performances and to guarantee a performance rate which is satisfactory regarding end-user requirements. This platform is composed of three main components: the library of programs, the knowledge base and the control component. The knowledge is either given by experts or learnt by the system. The control is generic in the sense that it is independent of any application. To validate this platform, we have built and evaluated six video surveillance systems which are featured with three properties: adaptability, reliability and real-time processing.