Object tracking via non-Euclidean geometry : a Grassmann approach

A robust visual tracking system requires an object appearance model that is able to handle occlusion, pose, and illumination variations in the video stream. This can be difficult to accomplish when the model is trained using only a single image. In this paper, we first propose a tracking approach based on affine subspaces (constructed from several images) which are able to accommodate the abovementioned variations. We use affine subspaces not only to represent the object, but also the candidate areas that the object may occupy. We furthermore propose a novel approach to measure affine subspace-to-subspace distance via the use of non-Euclidean geometry of Grassmann manifolds. The tracking problem is then considered as an inference task in a Markov Chain Monte Carlo framework via particle filtering. Quantitative evaluation on challenging video sequences indicates that the proposed approach obtains considerably better performance than several recent state-of-the-art methods such as Tracking-Learning-Detection and MILtrack.

Random projections on manifolds of symmetric positive definite matrices for image classification

Recent advances suggest that encoding images through Symmetric Positive Definite (SPD) matrices and then interpreting such matrices as points on Riemannian manifolds can lead to increased classification performance. Taking into account manifold geometry is typically done via (1) embedding the manifolds in tangent spaces, or (2) embedding into Reproducing Kernel Hilbert Spaces (RKHS). While embedding into tangent spaces allows the use of existing Euclidean-based learning algorithms, manifold shape is only approximated which can cause loss of discriminatory information. The RKHS approach retains more of the manifold structure, but may require non-trivial effort to kernelise Euclidean-based learning algorithms. In contrast to the above approaches, in this paper we offer a novel solution that allows SPD matrices to be used with unmodified Euclidean-based learning algorithms, with the true manifold shape well-preserved. Specifically, we propose to project SPD matrices using a set of random projection hyperplanes over RKHS into a random projection space, which leads to representing each matrix as a vector of projection coefficients. Experiments on face recognition, person re-identification and texture classification show that the proposed approach outperforms several recent methods, such as Tensor Sparse Coding, Histogram Plus Epitome, Riemannian Locality Preserving Projection and Relational Divergence Classification.

Summarisation of short-term and long-term videos using texture and colour

We present a novel approach to video summarisation that makes use of a Bag-of-visual-Textures (BoT) approach. Two systems are proposed, one based solely on the BoT approach and another which exploits both colour information and BoT features. On 50 short-term videos from the Open Video Project we show that our BoT and fusion systems both achieve state-of-the-art performance, obtaining an average F-measure of 0.83 and 0.86 respectively, a relative improvement of 9% and 13% when compared to the previous state-of-the-art. When applied to a new underwater surveillance dataset containing 33 long-term videos, the proposed system reduces the amount of footage by a factor of 27, with only minor degradation in the information content. This order of magnitude reduction in video data represents significant savings in terms of time and potential labour cost when manually reviewing such footage.

Matching image sets via adaptive multi convex hull

Traditional nearest points methods use all the samples in an image set to construct a single convex or affine hull model for classification. However, strong artificial features and noisy data may be generated from combinations of training samples when significant intra-class variations and/or noise occur in the image set. Existing multi-model approaches extract local models by clustering each image set individually only once, with fixed clusters used for matching with various image sets. This may not be optimal for discrimination, as undesirable environmental conditions (eg. illumination and pose variations) may result in the two closest clusters representing different characteristics of an object (eg. frontal face being compared to non-frontal face). To address the above problem, we propose a novel approach to enhance nearest points based methods by integrating affine/convex hull classification with an adapted multi-model approach. We first extract multiple local convex hulls from a query image set via maximum margin clustering to diminish the artificial variations and constrain the noise in local convex hulls. We then propose adaptive reference clustering (ARC) to constrain the clustering of each gallery image set by forcing the clusters to have resemblance to the clusters in the query image set. By applying ARC, noisy clusters in the query set can be discarded. Experiments on Honda, MoBo and ETH-80 datasets show that the proposed method outperforms single model approaches and other recent techniques, such as Sparse Approximated Nearest Points, Mutual Subspace Method and Manifold Discriminant Analysis.

Automatic classification of human epithelial type 2 cell indirect immunofluorescence images using cell pyramid matching

This paper describes a novel system for automatic classification of images obtained from Anti-Nuclear Antibody (ANA) pathology tests on Human Epithelial type 2 (HEp-2) cells using the Indirect Immunofluorescence (IIF) protocol. The IIF protocol on HEp-2 cells has been the hallmark method to identify the presence of ANAs, due to its high sensitivity and the large range of antigens that can be detected. However, it suffers from numerous shortcomings, such as being subjective as well as time and labour intensive. Computer Aided Diagnostic (CAD) systems have been developed to address these problems, which automatically classify a HEp-2 cell image into one of its known patterns (eg. speckled, homogeneous). Most of the existing CAD systems use handpicked features to represent a HEp-2 cell image, which may only work in limited scenarios. We propose a novel automatic cell image classification method termed Cell Pyramid Matching (CPM), which is comprised of regional histograms of visual words coupled with the Multiple Kernel Learning framework. We present a study of several variations of generating histograms and show the efficacy of the system on two publicly available datasets: the ICPR HEp-2 cell classification contest dataset and the SNPHEp-2 dataset.

Dictionary learning and sparse coding on Grassmann manifolds : an extrinsic solution

Recent advances in computer vision and machine learning suggest that a wide range of problems can be addressed more appropriately by considering non-Euclidean geometry. In this paper we explore sparse dictionary learning over the space of linear subspaces, which form Riemannian structures known as Grassmann manifolds. To this end, we propose to embed Grassmann manifolds into the space of symmetric matrices by an isometric mapping, which enables us to devise a closed-form solution for updating a Grassmann dictionary, atom by atom. Furthermore, to handle non-linearity in data, we propose a kernelised version of the dictionary learning algorithm. Experiments on several classification tasks (face recognition, action recognition, dynamic texture classification) show that the proposed approach achieves considerable improvements in discrimination accuracy, in comparison to state-of-the-art methods such as kernelised Affine Hull Method and graph-embedding Grassmann discriminant analysis.

Multi-shot person re-identification via relational stein divergence

Person re-identification is particularly challenging due to significant appearance changes across separate camera views. In order to re-identify people, a representative human signature should effectively handle differences in illumination, pose and camera parameters. While general appearance-based methods are modelled in Euclidean spaces, it has been argued that some applications in image and video analysis are better modelled via non-Euclidean manifold geometry. To this end, recent approaches represent images as covariance matrices, and interpret such matrices as points on Riemannian manifolds. As direct classification on such manifolds can be difficult, in this paper we propose to represent each manifold point as a vector of similarities to class representers, via a recently introduced form of Bregman matrix divergence known as the Stein divergence. This is followed by using a discriminative mapping of similarity vectors for final classification. The use of similarity vectors is in contrast to the traditional approach of embedding manifolds into tangent spaces, which can suffer from representing the manifold structure inaccurately. Comparative evaluations on benchmark ETHZ and iLIDS datasets for the person re-identification task show that the proposed approach obtains better performance than recent techniques such as Histogram Plus Epitome, Partial Least Squares, and Symmetry-Driven Accumulation of Local Features.

Improved image set classification via joint sparse approximated nearest subspaces

Existing multi-model approaches for image set classification extract local models by clustering each image set individually only once, with fixed clusters used for matching with other image sets. However, this may result in the two closest clusters to represent different characteristics of an object, due to different undesirable environmental conditions (such as variations in illumination and pose). To address this problem, we propose to constrain the clustering of each query image set by forcing the clusters to have resemblance to the clusters in the gallery image sets. We first define a Frobenius norm distance between subspaces over Grassmann manifolds based on reconstruction error. We then extract local linear subspaces from a gallery image set via sparse representation. For each local linear subspace, we adaptively construct the corresponding closest subspace from the samples of a probe image set by joint sparse representation. We show that by minimising the sparse representation reconstruction error, we approach the nearest point on a Grassmann manifold. Experiments on Honda, ETH-80 and Cambridge-Gesture datasets show that the proposed method consistently outperforms several other recent techniques, such as Affine Hull based Image Set Distance (AHISD), Sparse Approximated Nearest Points (SANP) and Manifold Discriminant Analysis (MDA).

Calculate excess mortality during heatwaves using Hilbert-Huang transform algorithm

Background Heatwaves could cause the population excess death numbers to be ranged from tens to thousands within a couple of weeks in a local area. An excess mortality due to a special event (e.g., a heatwave or an epidemic outbreak) is estimated by subtracting the mortality figure under ‘normal’ conditions from the historical daily mortality records. The calculation of the excess mortality is a scientific challenge because of the stochastic temporal pattern of the daily mortality data which is characterised by (a) the long-term changing mean levels (i.e., non-stationarity); (b) the non-linear temperature-mortality association. The Hilbert-Huang Transform (HHT) algorithm is a novel method originally developed for analysing the non-linear and non-stationary time series data in the field of signal processing, however, it has not been applied in public health research. This paper aimed to demonstrate the applicability and strength of the HHT algorithm in analysing health data. Methods Special R functions were developed to implement the HHT algorithm to decompose the daily mortality time series into trend and non-trend components in terms of the underlying physical mechanism. The excess mortality is calculated directly from the resulting non-trend component series. Results The Brisbane (Queensland, Australia) and the Chicago (United States) daily mortality time series data were utilized for calculating the excess mortality associated with heatwaves. The HHT algorithm estimated 62 excess deaths related to the February 2004 Brisbane heatwave. To calculate the excess mortality associated with the July 1995 Chicago heatwave, the HHT algorithm needed to handle the mode mixing issue. The HHT algorithm estimated 510 excess deaths for the 1995 Chicago heatwave event. To exemplify potential applications, the HHT decomposition results were used as the input data for a subsequent regression analysis, using the Brisbane data, to investigate the association between excess mortality and different risk factors. Conclusions The HHT algorithm is a novel and powerful analytical tool in time series data analysis. It has a real potential to have a wide range of applications in public health research because of its ability to decompose a nonlinear and non-stationary time series into trend and non-trend components consistently and efficiently.

Compressed sensing using hidden Markov models with application to vision based aircraft tracking

This paper investigates compressed sensing using hidden Markov models (HMMs) and hence provides an extension of recent single frame, bounded error sparse decoding problems into a class of sparse estimation problems containing both temporal evolution and stochastic aspects. This paper presents two optimal estimators for compressed HMMs. The impact of measurement compression on HMM filtering performance is experimentally examined in the context of an important image based aircraft target tracking application. Surprisingly, tracking of dim small-sized targets (as small as 5-10 pixels, with local detectability/SNR as low as − 1.05 dB) was only mildly impacted by compressed sensing down to 15% of original image size.

A pilot study on affective classification of facial images for emerging news topics

The proliferation of news reports published in online websites and news information sharing among social media users necessitates effective techniques for analysing the image, text and video data related to news topics. This paper presents the first study to classify affective facial images on emerging news topics. The proposed system dynamically monitors and selects the current hot (of great interest) news topics with strong affective interestingness using textual keywords in news articles and social media discussions. Images from the selected hot topics are extracted and classified into three categorized emotions, positive, neutral and negative, based on facial expressions of subjects in the images. Performance evaluations on two facial image datasets collected from real-world resources demonstrate the applicability and effectiveness of the proposed system in affective classification of facial images in news reports. Facial expression shows high consistency with the affective textual content in news reports for positive emotion, while only low correlation has been observed for neutral and negative. The system can be directly used for applications, such as assisting editors in choosing photos with a proper affective semantic for a certain topic during news report preparation.

Reversible and semi-blind relational database watermarking

Database watermarking has received significant research attention in the current decade. Although, almost all watermarking models have been either irreversible (the original relation cannot be restored from the watermarked relation) and/or non-blind (requiring original relation to detect the watermark in watermarked relation). This model has several disadvantages over reversible and blind watermarking (requiring only watermarked relation and secret key from which the watermark is detected and original relation is restored) including inability to identify rightful owner in case of successful secondary watermarking, inability to revert the relation to original data set (required in high precision industries) and requirement to store unmarked relation at a secure secondary storage. To overcome these problems, we propose a watermarking scheme that is reversible as well as blind. We utilize difference expansion on integers to achieve reversibility. The major advantages provided by our scheme are reversibility to high quality original data set, rightful owner identification, resistance against secondary watermarking attacks, and no need to store original database at a secure secondary storage.

Robust and reversible numerical set watermarking

Numeric sets can be used to store and distribute important information such as currency exchange rates and stock forecasts. It is useful to watermark such data for proving ownership in case of illegal distribution by someone. This paper analyzes the numerical set watermarking model presented by Sion et. al in “On watermarking numeric sets”, identifies it’s weaknesses, and proposes a novel scheme that overcomes these problems. One of the weaknesses of Sion’s watermarking scheme is the requirement to have a normally-distributed set, which is not true for many numeric sets such as forecast figures. Experiments indicate that the scheme is also susceptible to subset addition and secondary watermarking attacks. The watermarking model we propose can be used for numeric sets with arbitrary distribution. Theoretical analysis and experimental results show that the scheme is strongly resilient against sorting, subset selection, subset addition, distortion, and secondary watermarking attacks.

Endurance test-rig for diagnostics and prognostics of rolling element bearings

Detection of faults in roller element bearing is a topic widely discussed in the scientific field. Bearings diagnostics is usually performed by analyzing experimental signals, almost always vibration signals, measured during operation. A number of signal processing techniques have been proposed and applied to measured vibrations. The diagnostic effectiveness of the techniques depends on their capacities and on the environmental conditions (i.e. environmental noise). The current trend, especially from an industrial point of view, is to couple the prognostics to the diagnostics. The realization of a prognostic procedure require the definition of parameters able to describe the bearing condition during its operation. Monitoring the values of these parameters during time allows to define their trends depending on the progress of the wear. In this way, a relation between the variation of the selected parameters and the wear progress, useful for diagnostics and prognostics of bearings in real industrial applications, can be established. In this paper, a laboratory test-rig designed to perform endurance tests on roller element bearing is presented. Since the test-rig has operated for a short time, only some preliminary available results are discussed.

An experimental based assessment of the deviation of the bearing characteristic frequencies

Slippage in the contact roller-races has always played a central role in the field of diagnostics of rolling element bearings. Due to this phenomenon, vibrations triggered by a localized damage are not strictly periodic and therefore not detectable by means of common spectral functions as power spectral density or discrete Fourier transform. Due to the strong second order cyclostationary component, characterizing these signals, techniques such as cyclic coherence, its integrated form and square envelope spectrum have proven to be effective in a wide range of applications. An expert user can easily identify a damage and its location within the bearing components by looking for particular patterns of peaks in the output of the selected cyclostationary tool. These peaks will be found in the neighborhood of specific frequencies, that can be calculated in advance as functions of the geometrical features of the bearing itself. Unfortunately the non-periodicity of the vibration signal is not the only consequence of the slippage: often it also involves a displacement of the damage characteristic peaks from the theoretically expected frequencies. This issue becomes particularly important in the attempt to develop highly automated algorithms for bearing damage recognition, and, in order to correctly set thresholds and tolerances, a quantitative description of the magnitude of the above mentioned deviations is needed. This paper is aimed at identifying the dependency of the deviations on the different operating conditions. This has been possible thanks to an extended experimental activity performed on a full scale bearing test rig, able to reproduce realistically the operating and environmental conditions typical of an industrial high power electric motor and gearbox. The importance of load will be investigated in detail for different bearing damages. Finally some guidelines on how to cope with such deviations will be given, accordingly to the expertise obtained in the experimental activity.