Dian Gong

Dynamic Manifold Warping for view invariant action recognition

We address the problem of learning view-invariant 3D models of human motion from motion capture data, in order to recognize human actions from a monocular video sequence with arbitrary viewpoint. We propose a Spatio-Temporal Manifold (STM) model to analyze non-linear multivariate time series with latent spatial structure and apply it to recognize actions in the joint-trajectories space. Based on STM, a novel alignment algorithm Dynamic Manifold Warping (DMW) and a robust motion similarity metric are proposed for human action sequences, both in 2D and 3D. DMW extends previous works on spatio-temporal alignment by incorporating manifold learning. We evaluate and compare the approach to state-of-the-art methods on motion capture data and realistic videos. Experimental results demonstrate the effectiveness of our approach, which yields visually appealing alignment results, produces higher action recognition accuracy, and can recognize actions from arbitrary views with partial occlusion.

Structured Time Series Analysis for Human Action Segmentation and Recognition

We address the problem of structure learning of human motion in order to recognize actions from a continuous monocular motion sequence of an arbitrary person from an arbitrary viewpoint. Human motion sequences are represented by multivariate time series in the joint-trajectories space. Under this structured time series framework, we first propose Kernelized Temporal Cut (KTC), an extension of previous works on change-point detection by incorporating Hilbert space embedding of distributions, to handle the nonparametric and high dimensionality issues of human motions. Experimental results demonstrate the effectiveness of our approach, which yields realtime segmentation, and produces high action segmentation accuracy. Second, a spatio-temporal manifold framework is proposed to model the latent structure of time series data. Then an efficient spatio-temporal alignment algorithm Dynamic Manifold Warping (DMW) is proposed for multivariate time series to calculate motion similarity between action sequences (segments). Furthermore, by combining the temporal segmentation algorithm and the alignment algorithm, online human action recognition can be performed by associating a few labeled examples from motion capture data. The results on human motion capture data and 3D depth sensor data demonstrate the effectiveness of the proposed approach in automatically segmenting and recognizing motion sequences, and its ability to handle noisy and partially occluded data, in the transfer learning module.

Tracking Using Motion Patterns for Very Crowded Scenes

This paper proposes Motion Structure Tracker (MST) to solve the problem of tracking in very crowded structured scenes. It combines visual tracking, motion pattern learning and multi-target tracking. Tracking in crowded scenes is very challenging due to hundreds of similar objects, cluttered background, small object size, and occlusions. However, structured crowded scenes exhibit clear motion pattern(s), which provides rich prior information. In MST, tracking and detection are performed jointly, and motion pattern information is integrated in both steps to enforce scene structure constraint. MST is initially used to track a single target, and further extended to solve a simplified version of the multi-target tracking problem. Experiments are performed on real-world challenging sequences, and MST gives promising results. Our method significantly outperforms several state-of-the-art methods both in terms of track ratio and accuracy.

Kernelized temporal cut for online temporal segmentation and recognition

We address the problem of unsupervised online segmenting human motion sequences into different actions. Kernelized Temporal Cut (KTC), is proposed to sequentially cut the structured sequential data into different regimes. KTC extends previous works on online change-point detection by incorporating Hilbert space embedding of distributions to handle the nonparametric and high dimensionality issues. Based on KTC, a realtime online algorithm and a hierarchical extension are proposed for detecting both action transitions and cyclic motions at the same time. We evaluate and compare the approach to state-of-the-art methods on motion capture data, depth sensor data and videos. Experimental results demonstrate the effectiveness of our approach, which yields realtime segmentation, and produces higher action segmentation accuracy. Furthermore, by combining with sequence matching algorithms, we can online recognize actions of an arbitrary person from an arbitrary viewpoint, given realtime depth sensor input.

High Order Geometric Range Free Localization in Opportunistic Cognitive Sensor Networks

Position information of the primary user (PU) is important to the transmission among secondary users (SU) in cognitive sensor networks (CSN). A range free geometric localization algorithm aiming at collecting position information of PUs is proposed in this paper. Since PUs do not cooperate with SUs in CSN, opportunistic spectrum access (OSA) protocol is applied in this algorithm. Another difficulty in localization in CSN is the unavailability of received signal strength (RSS) or other precise measurements in the sensor nodes physics layer due to the power and size limitation of sensor nodes, so the algorithm in this paper is designed to be range free. This range free feature makes the algorithm robust against the uncertainty of parameters of the physics layer. With respect to the performance of the algorithm, an approximation of mean squared error (MSE) of localization is derived. The roles played by the parameters in the system are analyzed. Simulation results are also presented.

An Examination of Multivariate Time Series Hashing with Applications to Health Care

As large-scale multivariate time series data become increasingly common in application domains, such as health care and traffic analysis, researchers are challenged to build efficient tools to analyze it and provide useful insights. Similarity search, as a basic operator for many machine learning and data mining algorithms, has been extensively studied before, leading to several efficient solutions. However, similarity search for multivariate time series data is intrinsically challenging because (1) there is no conclusive agreement on what is a good similarity metric for multivariate time series data and (2) calculating similarity scores between two time series is often computationally expensive. In this paper, we address this problem by applying a generalized hashing framework, namely kernelized locality sensitive hashing, to accelerate time series similarity search with a series of representative similarity metrics. Experiment results on three large-scale clinical data sets demonstrate the effectiveness of the proposed approach.

Dynamic code assignment for OVSF code system

The OVSF-CDMA system must allocate user codes to support a new call, when it supports variable user data rates. Accordingly, the user code would be easily blocked without an efficient code reassignment algorithm. This paper presents a right first dynamic code assignment (RFDCA) algorithm, which aims at always keeping the OVSF codes tree in an optimal right first (RF) state on the basis of the number-topology count (Nt). This algorithm realizes a quick assignment of the OVSF codes to the new callers so that the reassignment of OVSF code could be minimized when new calls arrive, further improving the spectral efficiency of the systems. In this paper, we have analyzed the algorithm by adopting Markov process and have practiced the simulation experiment. The theoretical analysis and the simulation experiment acquire the same results, proving the effectiveness of the algorithm. More importantly, the result of the simulation experiment indicates the number of reassigned codes (NRC) is far fewer than that of other algorithms when new calls arrive

Probabilistic tensor voting for robust perceptual grouping

We address the problem of unsupervised segmentation and grouping in 2D and 3D space, where samples are corrupted by noise, and in the presence of outliers. The problem has attracted attention in previous research work, but non-parametric outlier filtering and inlier denoising are still challenging. Tensor voting is a non-parametric algorithm that can infer local data geometric structure. Standard tensor voting considers outlier noise explicitly, but may suffer from serious problems if the inlier data is also noisy. In this paper, we propose probabilistic Tensor Voting, a Bayesian extension of standard tensor voting, taking into consideration both probabilistic and geometric meaning. Probabilistic tensor voting explicitly considers both outlier and inlier noise, and can handle them simultaneously. In the new framework, the representation consists of a 2nd order symmetric tensor, a polarity vector, and a new type 2 polarity vector orthogonal to the first one. We give a theoretical interpretation of our framework. Experimental results show that our approach outperforms other methods, including standard tensor voting.

Dynamic System Analysis and Generalized Optimal Code Assignment of OVSF-CDMA Systems

OVSF-CDMA systems reallocate user codes when new calls arrive. To reduce code blocking due to new user codes, Generalized Optimal Right First Dynamic Code Assignment (GRFDCA) is brought up. GRFDCA keeps OVSF code trees in optimal right first (RF) states with respect to the number- topology (NT) count. Dynamic model of OVSF-CDMA systems is established. Tests of hypothesis about the Markovity of arrival processes are set up. Based on the Markovity assumption, continuous time Markov model of OVSF-CDMA systems is presented. By introducing absorbing states, phase type (PH) model is applied. Dynamic properties of OVSF-CDMA systems are analyzed based on PH model. Simulations and further discussions based on the dynamic model are presented.

Extracting micro-structural gabor features for face recognition

Robustness and discriminability are two key issues in face recognition. In this paper, we propose a new algorithm which extracts micro-structural Gabor feature to achieve good robustness and discriminability simultaneously. We first design a family of directional block partitions to compute the block-level directional projections of the classical Gabor feature. Then we use two statistical kernels, i.e, the mean kernel and the variance kernel, to extract the micro-structural statistics. Analysis of both robustness and discriminability is conducted to show that the new feature is not only more robust to misalignment, but also more discriminative than the classical down-sampling Gabor feature, which is further demonstrated by three groups of experiments on the BANCA dataset.