Search

Gait recognition can be used in person identification and re-identification by itself or in conjunction with other biometrics. Although gait has both spatial and temporal attributes, and it has been observed that decoupling spatial feature and temporal feature can better exploit the gait feature on the fine-grained level. However, the spatial-temporal correlations of gait video signals are also lost in the decoupling process. Direct 3D convolution approaches can retain such correlations, but they also introduce unnecessary interferences. Instead of common 3D convolution solutions, this paper proposes an integration of decoupling process into a 3D convolution framework for cross-view gait recognition. In particular, a novel block consisting of a Parallel-insight Convolution layer integrated with a Spatial-Temporal Dual-Attention (STDA) unit is proposed as the basic block for global spatial-temporal information extraction. Under the guidance of the STDA unit, this block can well integrate spatial-temporal information extracted by two decoupled models and at the same time retain the spatial-temporal correlations. In addition, a Multi-Scale Salient Feature Extractor is proposed to further exploit the fine-grained features through context awareness extension of part-based features and adaptively aggregating the spatial features. Extensive experiments on three popular gait datasets, namely CASIA-B, OULP and OUMVLP, demonstrate that the proposed method outperforms state-of-the-art methods.

International audience; Different from the traditional macro-expressions, micro-expressions are unconscious, quick and trustworthy facial expressions, which can reveal real emotion. Micro-expressions can provide information that is important and crucial in applications such as lie detection, criminal investigation, pain or mood assessment, etc. However, it is worth noting that most current micro-expression recognition methods rely only on a single micro-expression database. If the training and test samples belong to different domains, for example, different micro-expression databases, the accuracy of existing micro-expression recognition methods will decrease dramatically. To solve this problem, we propose an unsupervised cross-database micro-expression recognition method based on distribution adaptation. Compared with most advanced unsupervised cross-database recognition methods, the proposed method has better performance on micro-expression recognition tasks.

Most of the existing research focuses on the recognition of micro-expressions, and few studies how to recognize the action units of micro-expressions. This is due to the low intensity of the facial action unit, which is not easily to be recognized. To solve this problem, we proposed a micro-expression action unit recognition algorithm based on dynamic image and spatial pyramids. First, the video is passed through the dynamic image generation module to generate a dynamic image and extract the motion information contained in all frames. Then, given the subtle movement properties of micro-expressions, different levels of semantic features are obtained through spatial pyramids. It is also known that micro-expressions appear in the small range and are concentrated in local area of the face, so the regional feature network and attention mechanism are used for the image features of each layer. Finally, due to the weak correlation between each action unit, our models are trained separately. Experiments on CASME and CAS(ME)2 datasets verify that the proposed algorithm has shown better action unit recognition performance compared with other advanced methods.

Microexpression (ME) spotting is a crucial step for emotion analysis to detect people’s true emotions. However, the short duration, small motion amplitude, and limited sample number make accurate spotting and ME-locating challenging. To address these problems, we make two contributions in this article, including an ME dataset SDU2 and a spotting method. The dataset SDU2 contains hybrid expressions of 1602 video clips labeled by professional psychologists, covering six main categories of emotions with balanced distribution. Our ME spotting method is based on a magnitude- and angle-combined optical flow feature, exploiting the angle information, which has been overlooked by other spotting methods. In this method, we select four most discriminative regions of interest (ROIs) and determine the appearance of MEs using the proposed feature and a novel decision criterion based on the characteristic of the optical flow angle and magnitude. We have conducted experiments on SDU2 and CASME II datasets. The results demonstrate that our method achieves much better performance compared to other state-of-the-art methods in terms of spotting accuracy.

Long-term person re-identification (re-ID) aims to build identity correspondence of the Target Subject of Interest (TSI) exposed under surveillance cameras over a long time interval. Compared to the conventional short-term re-ID studied by most existing works, it suffers an additional problem: significant dressing change observed with time lapsing. Unfortunately, this variation in long-term person re-ID case contradicts the assumption of prior short-term re-ID approaches, and thus causes significant difficulties if conventional short-term re-ID methods are applied. To address the problem, this paper proposes to learn hybrid feature representation via a two-stream network named SpTSkM, including a spatial-temporal stream and a skeleton motion stream. The former performs directly on image sequences, which tends to learn identity-related spatial-temporal patterns such as body geometric structure and body movement. The latter operates on normalized 3D skeletons by adapting graph convolutional network, which tends to learn pure motion patterns from skeleton sequences. Both streams extract fine-grained level time-gap stable information that is robust to appearance changes in long-term re-ID and meanwhile maintains sufficient discriminability to differentiate different people. The final matching metric is obtained by mixing information of the two streams in a score-level fusion strategy. In addition, we collect a Cloth-Varying vIDeo re-ID (CVID-reID) dataset particularly for long-term re-ID. It contains video tracklets of celebrities posted on the Internet. These videos are snapshots under extremely different scenarios that include highly dynamic background, diverse camera views and abundant cloth variations on each TSI. These factors cause CVID-reID more complicated and closer to practice. Our experiments demonstrate the difficulty of long-term person re-ID and also validate the effectiveness of the proposed SpTSkM, showing the best performance.

Tensor analysis methods have played an important role in identifying human gaits using high dimensional data. However, when view angles change, it becomes more and more difficult to recognize cross-view gait by learning only a set of multi-linear projection matrices. To address this problem, a general tensor representation framework for cross-view gait recognition is proposed in this paper. There are three criteria of tensorial coupled mappings in the proposed framework. (1) Coupled multi-linear locality-preserved criterion (CMLP) aims to detect the essential tensorial manifold structure via preserving local information. (2) Coupled multi-linear marginal fisher criterion (CMMF) aims to encode the intra-class compactness and inter-class separability with local relationships. (3) Coupled multi-linear discriminant analysis criterion (CMDA) aims to minimize the intra-class scatter and maximize the inter-class scatter. For the three tensor algorithms for cross-view gaits, two sets of multi-linear projection matrices are iteratively learned using alternating projection optimization procedures. The proposed methods are compared with the recently published cross-view gait recognition approaches on CASIA(B) and OU-ISIR gait database. The results demonstrate that the performances of the proposed methods are superior to existing state-of-the-art cross-view gait recognition approaches.

Multiview representation has become important due to its good performance for machine learning problems. In this paper, a multiview representation framework based on transfer learning is proposed for micro-expression recognition. The framework takes macro-expression as the auxiliary domain and micro-expression as the target domain, and assists the identification of micro-expressions by transferring the rich information extracted from the auxiliary domain, which effectively addresses the small sample problem of micro-expression recognition. The proposed algorithm mainly consists of three parts. Firstly, the features of the two domains are projected into a common space and the dictionaries of each domain are studied respectively. Then the dictionary of micro-expression domain is linearly reconstructed. Finally, in order to improve the comprehensive utilization of feature information, the most representative features from four different micro-expression feature sets are selected by multiview representation. The experiments and evaluation are carried out on three different databases, and the performance comparison of the proposed algorithm with other advanced methods are given. The experimental results show that the proposed algorithm has the better performance than other related methods.

International audience; Unlike the conventional facial expressions, micro-expressions are involuntary and transient facial expressions capable of revealing the genuine emotions that people attempt to hide. Therefore, they can provide important information in a broad range of applications such as lie detection, criminal detection, etc. Since micro-expressions are transient and of low intensity, however, their detection and recognition is difficult and relies heavily on expert experiences. Due to its intrinsic particularity and complexity, video-based micro-expression analysis is attractive but challenging, and has recently become an active area of research. Although there have been numerous developments in this area, thus far there has been no comprehensive survey that provides researchers with a systematic overview of these developments with a unified evaluation. Accordingly, in this survey paper, we first highlight the key differences between macro- and micro-expressions, then use these differences to guide our research survey of video-based micro-expression analysis in a cascaded structure, encompassing the neuropsychological basis, datasets, features, spotting algorithms, recognition algorithms, applications and evaluation of state-of-the-art approaches. For each aspect, the basic techniques, advanced developments and major challenges are addressed and discussed. Furthermore, after considering the limitations of existing micro-expression datasets, we present and release a new dataset - called micro-and-macro expression warehouse (MMEW) - containing more video samples and more labeled emotion types. We then perform a unified comparison of representative methods on CAS(ME) for spotting, and on MMEW and SAMM for recognition, respectively. Finally, some potential future research directions are explored and outlined.

Nowadays, ground-coupled heat pump system (GCHP) becomes one of the most energy-efficient systems in heating, cooling and hot water supply. However, it remains challenging to accurately predict thermal energy conversion, and the numerical calculation methods are too complicated. First, according to seasonality, this paper analyzes four variables, including the power consumption of heat pump, the power consumption of system, the ratios of the heating capacity (or the refrigerating capacity) of heat pump to the operating powers of heat pump and to the total system, respectively. Then, heat transfer performance of GCHP by historical data and working parameters is predicted by using random forests algorithm based on autoregressive model and introducing working parameters. Finally, we conduct experiments on 360-months (30-years) data generated by GCHP software. Among them, the first 300 months of data are used for training the model, and the last 60 months of data are used for prediction. Benefitting from the working condition inputs it contained, our model achieves lower Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE) than Exponential Smoothing (ES), Autoregressive Model (AR), Autoregressive Moving Average Model (ARMA) and Auto-regressive Integrated Moving Average Model (ARIMA) without working condition inputs.

Close-up talking heads are among the most common and salient object in video contents, such as face-to-face conversations in social media, teleconferences, news broadcasting, talk shows, etc. Due to the high sensitivity of human visual system to faces, compression distortions in talking heads videos are highly visible and annoying. To address this problem, we present a novel deep convolutional neural network (DCNN) method for very low bit rate video reconstruction of talking heads. The key innovation is a new DCNN architecture that can exploit the audio-video correlations to repair compression defects in the face region. We further improve reconstruction quality by embedding into our DCNN the encoder information of the video compression standards and introducing a constraining projection module in the network. Extensive experiments demonstrate that the proposed DCNN method outperforms the existing state-of-the-art methods on videos of talking heads.

© 1991-2012 IEEE. A problem that hinders good performance of general gait recognition systems is that the appearance features of gaits are more affected-prone by views than identities, especially when the walking direction of the probe gait is different from the register gait. This problem cannot be solved by traditional projection learning methods because these methods can learn only one projection matrix, and thus for the same subject, it cannot transfer cross-view gait features into similar ones. This paper presents an innovative method to overcome this problem by aligning gait energy images (GEIs) across views with the coupled bilinear discriminant projection (CBDP). Specifically, the CBDP generates the aligned gait matrix features for two views with two sets of bilinear transformation matrices, so that the original GEIs' spatial structure information can be preserved. By iteratively maximizing the ratio of inter-class distance metric to intra-class distance metric, the CBDP can learn the optimal matrix subspace where the GEIs across views are aligned in both horizontal and vertical coordinates. Therefore, the CBDP is also able to avoid the under-sample problem. We also theoretically prove that the upper and lower bounds of the objective function sequence of the CBDP are both monotonically increasing, so the convergence of the CBDP is demonstrated. In the terms of accuracy, the comparative experiments on the CASIA (B) and OU-ISIR gait databases show that our method is superior to the state-of-the-art cross-view gait recognition methods. More impressively, encouraging performance is obtained by our method even in matching a lateral-view gait with a frontal-view gait.

The convolutional neural network (CNN) is an excellent method for image recognition. However, there are some problems in the construction process of CNN model, such as network structure setting depends on experience knowledge, network parameters selection is difficult, and there is a lack of relevance between network model and training data. To overcome the shortcomings of CNN model construction theory, in this paper, we develop a new construction approach, named adaptive deep CNN network model based on data-driven. In our method, we first set up the initial CNN model in a simple way, and the initial model only contains one feature map in the convolution layer and pooling layer. And then, the network is adaptively constructed by using the idea of learning parameters and expanding network. In network expansion, the convergence rate of CNN model is used as evaluation index of global expansion, and some global branches are added to the network model. After global expansion, the CNN is controlled to local expansion according to the recognition rate of cross validation samples. The local network learning is stopped until the recognition rate reaches the expected value. Finally, the adaptive incremental learning of network structure is realized by expanding some new branches for new samples. Experimental results on two benchmark face databases, CMU-PIE face database and MIT-CBCL face database, demonstrate the effectiveness of the proposed method.

In this paper, we propose three effective binary face descriptor learning methods, namely dual-cross patterns from three orthogonal planes (DCP-TOP), hot wheel patterns (HWP) and HWP-TOP for macro/micro-expression representation. We use feature selection to make the binary descriptors compact. Because of the limited labeled micro-expression samples, we leverage abundant labeled macro-expression and speech samples to train a more accurate classifier. Coupled metric learning algorithm is employed to model the shared features between micro-expression samples and macro-information. Smooth SVM (SSVM) is selected as a classifier to evaluate the performance of micro-expression recognition. Extensive experimental results show that our proposed methods yield the state-of-the-art classification accuracies on the CASMEII database.

International audience; As one of the most important forms of psychological behaviors, micro-expression can reveal the real emotion. However, the existing labeled training samples are limited to train a high performance model. To overcome this limit, in this paper we propose a macro-to-micro transformation model which enables to transfer macro-expression learning to micro-expression. Doing so improves the efficiency of the micro-expression features. For this purpose, LBP and LBP-TOP are used to extract macro-expression features and micro-expression features, respectively. Furthermore, feature selection is employed to reduce redundant features. Finally, singular value decomposition is employed to achieve macro-to-micro transformation model. The experimental evaluation based on the incorporated database including CK+ and CASME2 demonstrates that the proposed model achieves a competitive performance compared with the existing micro-expression recognition methods.

Micro-expression has raised increasing attention for analyzing human inner emotions. However, most micro-expression recognition methods are developed with specific feature representations and extraction methods, such as local binary pattern on three orthogonal planes (LBP-TOP) and optical flow. The performance in such micro-expression recognition models is not high due to the limited training samples and the unequal size of the sample category. To improve the performance, we present a novel algorithm, named coupled source domain targetized with updating tag vectors, and we apply it to the micro-expression recognition. This method leverages rich speech data to enhance micro-expression recognition by transferring learning from the speech to the micro-expression recognition. The method highlights are: it simultaneously projects micro-expression samples and speech samples into a common space, then minimizes the reconstruction error between the speech and micro-expression samples, with an updating tag vectors added in the reconstruction process. It performs recognition by using dictionary learning together with support vector machine (SVM). Experimental results on the CASIA Chinese emotional corpus and CASME II micro-expression database demonstrate the effectiveness of our method.

© 1992-2012 IEEE. Gait recognition has attracted growing attention in recent years, as the gait of humans has a strong discriminative ability even under low resolution at a distance. Unfortunately, the performance of gait recognition can be largely affected by view change. To address this problem, we propose a coupled patch alignment (CPA) algorithm that effectively matches a pair of gaits across different views. To realize CPA, we first build a certain amount of patches, and each of them is made up of a sample as well as its intra-class and inter-class nearest neighbors. Then, we design an objective function for each patch to balance the cross-view intra-class compactness and the cross-view inter-class separability. Finally, all the local-independent patches are combined to render a unified objective function. Theoretically, we show that the proposed CPA has a close relationship with canonical correlation analysis. Algorithmically, we extend CPA to 'multi-dimensional patch alignment' that can handle an arbitrary number of views. Comprehensive experiments on CASIA(B), USF, and OU-ISIR gait databases firmly demonstrate the effectiveness of our methods over other existing popular methods in terms of cross-view gait recognition.

© 2016 Elsevier B.V. Gait recognition degrades dramatically when gaits are captured from different directions or at different distances due to the low similarity between the registration and the query. This paper addresses the distance metric learning problem in matching between dual cross-domain gaits. Most existing distance metric learning algorithms are only able to match among a set of single domain gaits, but fail to measure the similarity of cross-domain gaits. Traditional gait recognition faces serious challenges, such as various low resolution images, which is caused by acquisition at different distances or different sampling devices, and various body shapes captured by different direction cameras. This paper presents a novel nonlinear coupled mappings (NCMs) algorithm to successfully match between the cross-domain gaits. The relationships within the training data as nodes in a graph are modeled in the kernel space and the constraint is designed to make the difference minimized between cross-domain gaits for an identical subject. Meanwhile, it makes the cross-domain gaits for different subjects disperse more separately with a large margin by using the supervised similarity matrix. Comprehensive experiments show that the proposed algorithm obtains higher accuracy than state-of-the-art algorithms.

Gaussian fields and harmonic functions (GFHF) and flexible manifold embedding (FME) provide effective means to label learning, by virtue of which we can evaluate labels of unknown samples (i.e. samples with unknown labels). When applied to face recognition, they are faced with the challenge that the face image varies with illuminations and facial expressions and poses. Moreover, in face recognition applications, available samples with known labels are almost always not sufficient. Thus it is hard to exploit GFHF or FME to achieve very satisfactory face recognition performance. In this paper, a novel FME algorithm is proposed for face recognition. Our work has two main contributions. Firstly, it devices a score fusion scheme to predict the label of the original unknown sample. Secondly, it obtains mirror images of all original face images and views both mirror images and original face images as available samples. The experimental results demonstrate that algorithm proposed in this paper can perform very well in face recognition. A novel FME algorithm is proposed for face recognition.The proposed algorithm devices a fusion scheme to predict the label of unknown sample.The proposed algorithm takes both mirror images and original face images as available samples.The experiment demonstrates that the algorithm can perform well in face recognition.

Nowadays, the ground-coupled heat pump (GCHP) systems have been recognized as one of the most energy-efficient systems for heating, cooling and hot water supply in both residential and commercial buildings. However, the heat transfer of ground heat exchanger (GHE) involves in large spatial scales, long time span and complex influential factors. We develop a data mining framework constructed by using 1998 experimental data to study the effects of 12 input variables composed of seven borehole parameters, two U-tube parameters, two ground parameters and one circulating liquid parameter to accurately predict the heat transfer performance of GHE for GCHP systems in 10 years. Hence, selecting a suitable input configuration to improve the energy efficiency has important sustainability benefits. The role of each of independent variable explaining the output variables is analyzed by partial least squares regression. Furthermore, support vector regression and M5 Model Tree are, respectively, used to predict the heat transfer performance. Extensive simulations show that we can predict the average quantity of heat exchanger, temperature of ground around GHE, inlet temperature of heat pump unit with very low level of error.

Micro-expressions are rapid involuntary facial expressions which reveal one's genuine emotions people trying to disguise. To the best of our knowledge, few works have been done in designing a micro-expression recognition visual platform. In this paper, we preliminarily study micro-expression recognition and subsequently develop a micro-expression visual platform that includes feature expression, dimension reduction as well as real-time video testing, etc. The platform leverages Gabor wavelet filter for expression feature extraction, principal components analysis (PCA) and linear discriminant analysis (LDA) for dimension reduction, and support vector machine (SVM) for expression classification. By using the trained model of the platform, we are able to test against micro-expression recognition. Experimental results show that the proposed scheme performs well on the CASME II database. Besides, it also works well on real-time expression recognition.

Electroencephalogram(EEG)analysis has been widely used in disease diagnosis.The EEG detection of the patients with epilepsy can be used to make judgments about patients’ conditions in time,which is of great practical value.Therefore,the techniques of automatic detection,diagnosis and classification of epileptic EEG signals are urgently needed.In order to realize fast and accurate automatic detection and classification of the EEG signals during the normal,interictal and ictal periods of epilepsy,we propose an automatic classification and recognition method which combines the Real Adaboost algorithm based on error-correcting output codes(ECOC)with a feature extraction method based on sample entropy(SampEn)and wavelet packet energy in this paper.In the present study,we used the sample entropy of input signals and the energy of some parts of frequency bands as features,and then we classified the extracted features with the method combining ECOC with Real AdaBoost algorithm.In order to test the validity,we used the epilepsy database from the University of Bonn.The database has 5groups of EEG signals,which contains the data of normal people with their eyes open or closed,the data collected inside and outside of the epileptic foci from patients during their interictal period and the data from patients during their ictal period.The results showed that the method had strong abilities of classification and recognition of the EEG signals,and especially the recognition rate had been improved significantly.The average recognition rate of the EEG signals with different features during the three periods of the five groups mentioned above can reach 96.78%,which is superior to those with algorithms recorded in many other literatures.The method has better stability,processing speed and potential of real-time application,and it plays a supporting role in the prediction and detection of epilepsy in clinical practice.

We contribute, through this paper, to design a novel algorithm called maximum margin projection with tensor representation (MMPTR). This algorithm is able to recognize gait and micro-expression represented as third-order tensors. Through maximizing the inter-class Laplacian scatter and minimizing the intra-class Laplacian scatter, MMPTR can seek a tensor-to-tensor projection that directly extracts discriminative and geometry-preserving features from the original tensorial data. We show the validity of MMPTR through extensive experiments on the CASIA(B) gait database, TUM GAID gait database, and CASME micro-expression database. The proposed MMPTR generally obtains higher accuracy than MPCA, GTDA as well as state-of-the-art DTSA algorithm. Experimental results included in this paper suggest that MMPTR is especially effective in such tensorial object recognition tasks.

This study, for the first time, developed an adaptive neural networks (NNs) formulation for the two-dimensional principal component analysis (2DPCA), whose space complexity is far lower than that of its statistical version. Unlike the NNs formulation of principal component analysis (PCA, i.e., 1DPCA), the solution with lower iteration in nature aims to directly deal with original image matrices. We also put forward the consistence in the conceptions of `eigenfaces' or `eigengaits' in both 1DPCA and 2DPCA neural networks. To evaluate the performance of the proposed NN, the experiments were carried out on AR face database and on 64 × 64 pixels gait energy images on CASIA(B) gait database. The less reconstruction error was exploited using the proposed NN in the condition of a large sample set compared to adaptive estimation of learning algorithms for NNs of PCA. On the contrary, if the sample set was small, the proposed NN could achieve a higher residue error than PCA NNs. The amount of calculation for the proposed NN here could be smaller than that for the PCA NNs on the feature extraction of the same image matrix, which represented an efficient solution to the problem of training images directly. On face and gait recognition tasks, a simple nearest neighbor classifier test indicated a particular benefit of the neural network developed here which serves as an efficient alternative to conventional PCA NNs.

Canonical correlation analysis (CCA) is a kind of classical multivariate analysis method. Less canonical correlation variables are used to describe the relationship between two variables completely but easily. To get high face recognition rate under low-resolution degradation over a long distance solidly, in this work, CCA is used to extract the correlation between high-resolution face images and low-resolution ones and to find the transform pair between them. Therefore, face images of the same individual with variable resolutions can be matched accurately. This is the first method that uses CCA to do low-resolution degradation face recognition over long distances. We conduct the experiments on the Extended Yale B and ORL database, and the experimental results validate the efficacy of the proposed method.

This paper introduces an orthogonal multilinear discriminant analysis (OMDA) algorithm for gait recognition. The discriminant feature vectors of OMDA are orthogonal to each other. With the advantage of extracting a portion of local information and reducing computational complexity, the subblock tensor analysis is employed to OMDA, named subblock orthogonal multilinear discriminant analysis (SOMDA). Considering that the vectors from different subblocks have different contributions to recognition, these vectors are given different weights and synthesized into a whole vector in the recognition process. We have conducted a comparative study on gait recognition to evaluate OMDA and SOMDA in terms of classification. With the tensor vectorization methods according to both variance and class discriminability, the OMDA-based recognition algorithm indicates that it outperforms other multilinear subspace solutions such as MPCA, MPCA + LDA, GTDA, DATER and UMDA. In the subblock experiments, it indicates that SOMDA is an improvement over OMDA.

This paper presents a new method for facial expression recognition using divided local binary pattern features. Motivated by Fourier’s parity decomposition, we decompose a LBP descriptor into two subdescriptors. This method reduces the amount of data of the feature extraction, reducing the data dimension effectively. Experimental results on JAFFE facial expression database show that the proposed method can get significant performance improvement in recognition rate, reduce the recognition time greatly.

The performances of biometrics may be adversely impact by different walking states, walking directions, resolutions of gait sequence images, pose variation and low resolution of face images. To address these problems, we presented a kernel coupled distance metric learning (KCDML) method after considering matching among different data collections. By using a kernel trick and a specialized locality preserving criterion, we formulated the problem of kernel coupled distance metric learning as an optimization problem whose aims are to search for the pair-wise samples staying as close as possible and to preserve the local structure intrinsic data geometry. Instead of an iterative solution, one single generalized eigen-decomposition can be leveraged to compute the two transformation matrices for two classifications of data sets. The effectiveness of the proposed method is empirically demonstrated on gait and face recognition tasks' results which outperform four linear subspace solutions' (i.e. CDML, PCA, LPP, LDA) and four nonlinear subspace solutions' (i.e. Huang's method, PCA-RBF, KPCA, KLPP).

A biometrics technique based on metric learning approach is proposed in this paper to achieve higher correct classification rates under the condition that the feature of the query is very different from that of the register for a given individual. Inspired by the definition of generalized distance, the criterion of this new metric learning is defined by finding an embedding that preserves local information and obtaining a subspace that best detects the essential manifold structure. Furthermore, the two transformation matrices for the query and the register are obtained by a generalized eigen-decomposition. Experiments tested on biometric applications of CASIA(B) gait database and the UMIST face database, demonstrate that our proposed method performs better than classical metric learning methods and the current radial basis function (RBF) algorithms.

A new gait period detection algorithm, dual-ellipse fitting (DEF) approach, is proposed. DEF is that two regions of the whole silhouette divided by the centroid are fitted into two ellipses, respectively. We construct the gait fluctuation as a periodic function which depends on the eccentricities of two halves of the silhouette over time. Experimental results show that the proposed method is robust to scale, translation, direction of walking and carrying a bag.

An innovative method for cascade eye detection using facial geometric features is proposed. The cascade detector is used to detect pairs of eye of candidates. The size similarity, the horizontal symmetry, the horizontal sextant angle, and the scale coefficients extracted from facial geometric feature are integrated to improve eye detection. Both the BioID database and real-time video stream were used in the experiments. Experimental results show that the proposed eye detection method is significantly more accurate than the cascade detector alone without sacrificing overmuch time-efficiency.