A coupled discriminative dictionary and transformation learning approach with applications to cross domain matching

Cross domain and cross-modal matching has many applications in the field of computer vision and pattern recognition. A few examples are heterogeneous face recognition, cross view action recognition, etc. This is a very challenging task since the data in two domains can differ significantly. In this work, we propose a coupled dictionary and transformation learning approach that models the relationship between the data in both domains. The approach learns a pair of transformation matrices that map the data in the two domains in such a manner that they share common sparse representations with respect to their own dictionaries in the transformed space. The dictionaries for the two domains are learnt in a coupled manner with an additional discriminative term to ensure improved recognition performance. The dictionaries and the transformation matrices are jointly updated in an iterative manner. The applicability of the proposed approach is illustrated by evaluating its performance on different challenging tasks: face recognition across pose, illumination and resolution, heterogeneous face recognition and cross view action recognition. Extensive experiments on five datasets namely, CMU-PIE, Multi-PIE, ChokePoint, HFB and IXMAS datasets and comparisons with several state-of-the-art approaches show the effectiveness of the proposed approach. (C) 2015 Elsevier B.V. All rights reserved.

A coupled discriminative dictionary and transformation learning approach with applications to cross domain matching

Cross domain and cross-modal matching has many applications in the field of computer vision and pattern recognition. A few examples are heterogeneous face recognition, cross view action recognition, etc. This is a very challenging task since the data in two domains can differ significantly. In this work, we propose a coupled dictionary and transformation learning approach that models the relationship between the data in both domains. The approach learns a pair of transformation matrices that map the data in the two domains in such a manner that they share common sparse representations with respect to their own dictionaries in the transformed space. The dictionaries for the two domains are learnt in a coupled manner with an additional discriminative term to ensure improved recognition performance. The dictionaries and the transformation matrices are jointly updated in an iterative manner. The applicability of the proposed approach is illustrated by evaluating its performance on different challenging tasks: face recognition across pose, illumination and resolution, heterogeneous face recognition and cross view action recognition. Extensive experiments on five datasets namely, CMU-PIE, Multi-PIE, ChokePoint, HFB and IXMAS datasets and comparisons with several state-of-the-art approaches show the effectiveness of the proposed approach. (C) 2015 Elsevier B.V. All rights reserved.

Anomalous reduction in domain wall displacement at the morphotropic phase boundary of the piezoelectric alloy system PbTiO3-BiScO3

A comparative study of field-induced domain switching and lattice strain was carried out by in situ electric-field-dependent high-energy synchrotron x-ray diffraction on a morphotropic phase boundary (MPB) and a near-MPB rhombohedral/pseudomonoclinic composition of a high-performance piezoelectric alloy (1-x) PbTiO3-(x)BiScO3. It is demonstrated that the MPB composition showing large d(33) similar to 425 pC/N exhibits significantly reduced propensity of field-induced domain switching as compared to the non-MPB rhombohedral composition (d(33) similar to 260 pC/N). These experimental observations contradict the basic premise of the martensitic-theory-based explanation which emphasizes on enhanced domain wall motion as the primary factor for the anomalous piezoelectric response in MPB piezoelectrics. Our results favor field-induced structural transformation to be the primary mechanism contributing to the large piezoresponse of the critical MPB composition of this system.

CROWD FLOW SEGMENTATION IN COMPRESSED DOMAIN USING CRF

Crowd flow segmentation is an important step in many video surveillance tasks. In this work, we propose an algorithm for segmenting flows in H.264 compressed videos in a completely unsupervised manner. Our algorithm works on motion vectors which can be obtained by partially decoding the compressed video without extracting any additional features. Our approach is based on modelling the motion vector field as a Conditional Random Field (CRF) and obtaining oriented motion segments by finding the optimal labelling which minimises the global energy of CRF. These oriented motion segments are recursively merged based on gradient across their boundaries to obtain the final flow segments. This work in compressed domain can be easily extended to pixel domain by substituting motion vectors with motion based features like optical flow. The proposed algorithm is experimentally evaluated on a standard crowd flow dataset and its superior performance in both accuracy and computational time are demonstrated through quantitative results.

ON THE UTILITY OF CANONICAL CORRELATION ANALYSIS FOR DOMAIN ADAPTATION IN MULTI-VIEW HEADPOSE ESTIMATION

The utility of canonical correlation analysis (CCA) for domain adaptation (DA) in the context of multi-view head pose estimation is examined in this work. We consider the three problems studied in 1], where different DA approaches are explored to transfer head pose-related knowledge from an extensively labeled source dataset to a sparsely labeled target set, whose attributes are vastly different from the source. CCA is found to benefit DA for all the three problems, and the use of a covariance profile-based diagonality score (DS) also improves classification performance with respect to a nearest neighbor (NN) classifier.

Magnetic properties and magnetic domain structure of bulk glass forming Nd60Al10Fe20Co10 alloy

The transition from hard to soft magnetic behaviour with increasing quenching rate is shown for Nd60WAl10Fe20Co10 melt-spun ribbons with different thickness. Microstructure and magnetic domain structure of ribbons were studied by magnetic force microscopy (MFM). Particle sizes < 5 nm decreasing gradually with increasing quenching rate were deduced from topographic images which differ from large-scale magnetic domains with a periodicity of about 350 nm in all ribbons irrespective the coercivity. This indicates that the magnetic properties of the alloy are governed by interaction of small magnetic particles. It is concluded that the presence of short-range-ordered structures with a local ordering similar to the Al metastable Nd-Fe binary phase is responsible for the hard magnetic properties in samples subjected to relatively low quenching rate.

Domain structure of hard magnetic NdAlFeCo bulk metallic glass

Magnetic domain structure of hard magnetic Nd60Al10Fe20Co10 bulk metallic glass (BMG) has been studied by using magnetic force microscopy. In the magnetic force images it is shown that the exchange interaction type magnetic domains with a period of about 360 nm do exist in the BMG, which is believed to be associated with the appearance of hard-magnetic properties in this system. As the scale of the magnetic domain is much larger than the size of the short-range ordered atomic clusters existing in the BMG, it is believed that the large areas of magnetic contrast are actually a collection of a group of clusters aligned in parallel by strong exchange coupling interaction. After fully crystallization, the BMG exhibits paramagnetism. No obvious magnetic contrast is observed in the magnetic force images of fully crystallized samples, except for a small quantity of ferromagnetic crystalline phase with low coercivity and an average size of 900 nm.

Domain structure of a Nd60Al10Fe20Co10 bulk metallic glass

Magnetic domain structure of Nd60Al10Fe20Co10 bulk metallic glass (BMG) has been studied by using magnetic-force microscopy. In the magnetic-force images it is shown that the exchange-interaction-type magnetic domains with a period of about 360 nm do exist in the BMG, which is believed to be associated with the appearance of hard-magnetic properties in this system. The existence of the large-scale domains demonstrates that the magnetic moments of a great deal of short-scale ordered atomic clusters in the BMG have been aligned by exchange coupling. Annealing at 715 K leads to partial crystallization of the BMG. However, the exchange coupling is stronger in the annealed sample, which is considered to arise from the increase of transition-metal concentration in the amorphous phase due to the precipitation of Nd crystalline phase.

Flux closure vortexlike domain structures in ferroelectric thin films

Enhanced piezoresponse force microscopy was used to study flux closure vortexlike structures of 90° ferroelastic domains at the nanoscale in thin ferroelectric lead zirconium titanate (PZT) films. Using an external electric field, a vortexlike structure was induced far away from a grain boundary, indicating that physical edges are not necessary for nucleation contrary to previous suggestions. We demonstrate two different configurations of vortexlike structures, one of which has not been observed before. The stability of these structures is found to be size dependent, supporting previous predictions. © 2010 The American Physical Society.

High peak power femtosecond pulse VECSELs for terahertz time domain spectroscopy

We report on a high peak power femtosecond modelocked VECSEL and its application as a drive laser for an all semiconductor terahertz time domain spectrometer. The VECSEL produced near-transform-limited 335 fs sech2 pulses at a fundamental repetition rate of 1 GHz, a centre wavelength of 999 nm and an average output power of 120 mW. We report on the effect that this high peak power and short pulse duration has on our generated THz signal.

A time domain computation method for dynamic behavior of mooring system

A quasi-steady time domain method is developed for the prediction of dynamic behavior of a mooring system under the environmental disturbances, such as regular or irregular waves, winds and currents. The mooring forces are obtained in a static sense at each instant. The dynamic feature of the mooring cables can be obtained by incorporating the extended 3-D lumped-mass method with the known ship motion history. Some nonlinear effects, such as the influence of the instantaneous change of the wetted hull surface on the hydrostatic restoring forces and Froude-Krylov forces, are included. The computational results show a satisfactory agreement with the experimental ones.

Energy principle of ferroelectric ceramics and single domain mechanical model

Many physical experiments have shown that the domain switching in a ferroelectric material is a complicated evolution process of the domain wall with the variation of stress and electric field. According to this mechanism, the volume fraction of the domain switching is introduced in the constitutive law of ferroelectric ceramic and used to study the nonlinear constitutive behavior of ferroelectric body in this paper. The principle of stationary total energy is put forward in which the basic unknown quantities are the displacement u (i) , electric displacement D (i) and volume fraction rho (I) of the domain switching for the variant I. Mechanical field equation and a new domain switching criterion are obtained from the principle of stationary total energy. The domain switching criterion proposed in this paper is an expansion and development of the energy criterion. On the basis of the domain switching criterion, a set of linear algebraic equations for the volume fraction rho (I) of domain switching is obtained, in which the coefficients of the linear algebraic equations only contain the unknown strain and electric fields. Then a single domain mechanical model is proposed in this paper. The poled ferroelectric specimen is considered as a transversely isotropic single domain. By using the partial experimental results, the hardening relation between the driving force of domain switching and the volume fraction of domain switching can be calibrated. Then the electromechanical response can be calculated on the basis of the calibrated hardening relation. The results involve the electric butterfly shaped curves of axial strain versus axial electric field, the hysteresis loops of electric displacement versus electric filed and the evolution process of the domain switching in the ferroelectric specimens under uniaxial coupled stress and electric field loading. The present theoretic prediction agrees reasonably with the experimental results given by Lynch.

Wave propagation and the frequency domain Green's functions in viscoelastic Biot/squirt (BISQ) media

In this paper, we examine the characteristics of elastic wave propagation in viscoelastic porous media, which contain simultaneously both the Biot-flow and the squirt-flow mechanisms (BISQ). The frequency-domain Green's functions for viscoelastic BISQ media are then derived based on the classic potential function methods. Our numerical results show that S-waves are only affected by viscoelasticity, but not by squirt-flows. However, the phase velocity and attenuation of fast P-waves are seriously influenced by both viscoelasticity and squirt-flows; and there exist two peaks in the attenuation-frequency variations of fast P-waves. In the low-frequency range, the squirt-flow characteristic length, not viscoelasticity, affects the phase velocity of slow P-waves, whereas it is opposite in the high-frequency range. As to the contribution of potential functions of two types of compressional waves to the Green's function, the squirt-flow length has a small effect, and the effects of viscoelastic parameter are mainly in the higher frequency range. Crown Copyright (C) 2006 Published by Elsevier Ltd. All rights reserved.