Towards view-invariant gait modeling: Computing view-normalized body part trajectories

This paper proposes an approach to compute view-normalized body part trajectories of pedestrians walking on potentially non-linear paths. The proposed approach finds applications in gait modeling, gait biometrics, and in medical gait analysis. Our approach uses the 2D trajectories of both feet and the head extracted from the tracked silhouettes. On that basis, it computes the apparent walking (sagittal) planes for each detected gait half-cycle. A homography transformation is then computed for each walking plane to make it appear as if walking was observed from a fronto-parallel view. Finally, each homography is applied to head and feet trajectories over each corresponding gait half-cycle. View normalization makes head and feet trajectories appear as if seen from a fronto-parallel viewpoint, which is assumed to be optimal for gait modeling purposes. The proposed approach is fully automatic as it requires neither manual initialization nor camera calibration. An extensive experimental evaluation of the proposed approach confirms the validity of the normalization process.     

Latent classification models for binary data

One of the simplest, and yet most consistently well-performing set of classifiers is the naïve Bayes models (a special class of Bayesian network models). However, these models rely on the (naïve) assumption that all the attributes used to describe an instance are conditionally independent given the class of that instance. To relax this independence assumption, we have in previous work proposed a family of models, called latent classification models (LCMs). LCMs are defined for continuous domains and generalize the naïve Bayes model by using latent variables to model class-conditional dependencies between the attributes. In addition to providing good classification accuracy, the LCM has several appealing properties, including a relatively small parameter space making it less susceptible to over-fitting. In this paper we take a first step towards generalizing LCMs to hybrid domains, by proposing an LCM for domains with binary attributes. We present algorithms for learning the proposed model, and we describe a variational approximation-based inference procedure. Finally, we empirically compare the accuracy of the proposed model to the accuracy of other classifiers for a number of different domains, including the problem of recognizing symbols in black and white images.     

Combining feature spaces for classification

In this paper we offer a variational Bayes approximation to the multinomial probit model for basis expansion and kernel combination. Our model is well-founded within a hierarchical Bayesian framework and is able to instructively combine available sources of information for multinomial classification. The proposed framework enables informative integration of possibly heterogeneous sources in a multitude of ways, from the simple summation of feature expansions to weighted product of kernels, and it is shown to match and in certain cases outperform the well-known ensemble learning approaches of combining individual classifiers. At the same time the approximation reduces considerably the CPU time and resources required with respect to both the ensemble learning methods and the full Markov chain Monte Carlo, Metropolis-Hastings within Gibbs solution of our model. We present our proposed framework together with extensive experimental studies on synthetic and benchmark datasets and also for the first time report a comparison between summation and product of individual kernels as possible different methods for constructing the composite kernel matrix.     

On the use of small training sets for neural network-based characterization of mixed pixels in remotely sensed hyperspectral images

In this work, neural network-based models involved in hyperspectral image spectra separation are considered. Focus is on how to select the most highly informative samples for effectively training the neural architecture. This issue is addressed here by several new algorithms for intelligent selection of training samples: (1) a border-training algorithm (BTA) which selects training samples located in the vicinity of the hyperplanes that can optimally separate the classes; (2) a mixed-signature algorithm (MSA) which selects the most spectrally mixed pixels in the hyperspectral data as training samples; and (3) a morphological-erosion algorithm (MEA) which incorporates spatial information (via mathematical morphology concepts) to select spectrally mixed training samples located in spatially homogeneous regions. These algorithms, along with other standard techniques based on orthogonal projections and a simple Maximin-distance algorithm, are used to train a multi-layer perceptron (MLP), selected in this work as a representative neural architecture for spectral mixture analysis. Experimental results are provided using both a database of nonlinear mixed spectra with absolute ground truth and a set of real hyperspectral images, collected at different altitudes by the digital airborne imaging spectrometer (DAIS 7915) and reflective optics system imaging spectrometer (ROSIS) operating simultaneously at multiple spatial resolutions.     

Fast, automatic and fine-grained tampered JPEG image detection via DCT coefficient analysis

The quick advance in image/video editing techniques has enabled people to synthesize realistic images/videos conveniently. Some legal issues may arise when a tampered image cannot be distinguished from a real one by visual examination. In this paper, we focus on JPEG images and propose detecting tampered images by examining the double quantization effect hidden among the discrete cosine transform (DCT) coefficients. To our knowledge, our approach is the only one to date that can automatically locate the tampered region, while it has several additional advantages: fine-grained detection at the scale of 8×8 DCT blocks, insensitivity to different kinds of forgery methods (such as alpha matting and inpainting, in addition to simple image cut/paste), the ability to work without fully decompressing the JPEG images, and the fast speed. Experimental results on JPEG images are promising.     

A new dual wing harmonium model for document retrieval

A new dual wing harmonium model that integrates term frequency features and term connection features into a low dimensional semantic space without increase of computation load is proposed for the application of document retrieval. Terms and vectorized graph connectionists are extracted from the graph representation of document by employing weighted feature extraction method. We then develop a new dual wing harmonium model projecting these multiple features into low dimensional latent topics with different probability distributions assumption. Contrastive divergence algorithm is used for efficient learning and inference. We perform extensive experimental verification, and the comparative results suggest that the proposed method is accurate and computationally efficient for document retrieval.     

Parameter estimation of Poisson mixture with automated model selection through BYY harmony learning

Finite mixture is widely used in the fields of information processing and data analysis. However, its model selection, i.e., the selection of components in the mixture for a given sample data set, has been still a rather difficult task. Recently, the Bayesian Ying-Yang (BYY) harmony learning has provided a new approach to the Gaussian mixture modeling with a favorite feature that model selection can be made automatically during parameter learning. In this paper, based on the same BYY harmony learning framework for finite mixture, we propose an adaptive gradient BYY learning algorithm for Poisson mixture with automated model selection. It is demonstrated well by the simulation experiments that this adaptive gradient BYY learning algorithm can automatically determine the number of actual Poisson components for a sample data set, with a good estimation of the parameters in the original or true mixture where the components are separated in a certain degree. Moreover, the adaptive gradient BYY learning algorithm is successfully applied to texture classification.     

3D noisy discrete objects: Segmentation and application to smoothing

We propose in this paper a segmentation process that can deal with noisy discrete objects. A flexible approach considering arithmetic discrete planes with a variable width is used to avoid the over-segmentation that might happen when classical segmentation algorithms based on regular discrete planes are used to decompose the surface of the object. A method to choose a seed and different segmentation strategies according to the shape of the surface is also proposed, as well as an application to smooth the border of convex noisy discrete objects.     

Stable local dimensionality reduction approaches

Dimensionality reduction is a big challenge in many areas. A large number of local approaches, stemming from statistics or geometry, have been developed. However, in practice these local approaches are often in lack of robustness, since in contrast to maximum variance unfolding (MVU), which explicitly unfolds the manifold, they merely characterize local geometry structure. Moreover, the eigenproblems that they encounter, are hard to solve. We propose a unified framework that explicitly unfolds the manifold and reformulate local approaches as the semi-definite programs instead of the above-mentioned eigenproblems. Three well-known algorithms, locally linear embedding (LLE), laplacian eigenmaps (LE) and local tangent space alignment (LTSA) are reinterpreted and improved within this framework. Several experiments are presented to demonstrate the potential of our framework and the improvements of these local algorithms.     

Computing upper and lower bounds of rotation angles from digital images

Rotations in the discrete plane are important for many applications such as image matching or construction of mosaic images. We suppose that a digital image A is transformed to another digital image B by a rotation. In the discrete plane, there are many angles giving the rotation from A to B, which we call admissible rotation angles from A to B. For such a set of admissible rotation angles, there exist two angles that achieve the lower and the upper bounds. To find those lower and upper bounds, we use hinge angles as used in Nouvel and Rémila [Incremental and transitive discrete rotations, in: R. Reulke, U. Eckardt, B. Flash, U. Knauer, K. Polthier (Eds.), Combinatorial Image Analysis, Lecture Notes in Computer Science, vol. 4040, Springer, Berlin, 2006, pp. 199-213]. A sequence of hinge angles is a set of particular angles determined by a digital image in the sense that any angle between two consecutive hinge angles gives the identical rotation of the digital image. We propose a method for obtaining the lower and the upper bounds of admissible rotation angles using hinge angles from a given Euclidean angle or from a pair of corresponding digital images.