Rashid Ansari

Pitch modification of speech using a low-sensitivity inverse filter approach

A simple and effective method for modifying the pitch of recorded speech units is described. This method was developed to overcome some limitations in the promising residual-excited linear prediction (RELP) technique. The key difference is that the choice of filter parameters in the new method is driven by a need for reducing sensitivity to pitch modification, rather than creating a residual with minimum energy as in RELP. Speech modifications using this method are superior in quality to those obtained with RELP, while at the same time being less sensitive than RELP to errors in pitch marking.

Multiple object tracking with kernel particle filter

A new particle filter, kernel particle filter (KPF), is proposed for visual tracking for multiple objects in image sequences. The KPF invokes kernels to form a continuous estimate of the posterior density function and allocates particles based on the gradient derived from the kernel density estimate. A data association technique is also proposed to resolve the motion correspondence ambiguities that arise when multiple objects are present. The data association technique introduces minimal amount of computation by making use of the intermediate results obtained in particle allocation. We show that KPF performs robust multiple object tracking with improved sampling efficiency.

Structure and design of two-channel filter banks derived from a triplet of halfband filters

A novel approach for the construction of two-channel one-dimensional (1-D) biorthogonal filter banks is described. This approach is intended to help overcome limitations in an otherwise effective set of recently proposed techniques while retaining the advantages of a simple design and feature-rich structure. The novelty of the method rests on the use of a triplet of halfband fillers, which are combined in a convenient form of the transfer function with adjustable parameters that provide the necessary flexibility in obtaining the desired response. Design procedures are described for the construction of the filter banks by casting the approximation problem in a form that easily lends itself to Remez exchange algorithm. In these design procedures, the analysis and synthesis filters can be either finite-duration impulse response, infinite duration impulse response, or a hybrid. Examples of design of 1-D filter banks and extension of the technique to two-dimensional diamond filter banks are presented. Finally, subband image-coding examples are given to demonstrate the advantage of the proposed filter-bank structure.

Robust audio watermarking using frequency-selective spread spectrum

A novel audio watermarking scheme based on frequency-selective spread spectrum (FSSS) technique is presented. Unlike most of the existing spread spectrum (SS) watermarking schemes that use the entire audible frequency range for watermark embedding, the proposed scheme randomly selects subband(s) signal(s) of the host audio signal for watermark embedding. The proposed FSSS scheme provides a natural mechanism to exploit the band-dependent frequency-masking characteristics of the human auditory system to ensure the fidelity of the host audio signal and the robustness of the embedded information. Key attributes of the proposed scheme include reduced host interference in watermark detection, better fidelity, secure embedding and improved multiple watermark embedding capability. To detect the embedded watermark, two blind watermark detection methods are examined, one based on normalised correlation and the other based on estimation correlation. Extensive simulation results are presented to analyse the performance of the proposed scheme for various signal manipulations and standard benchmark attacks. A comparison with the existing full-band SS-based schemes is also provided to show the improved performance of the proposed scheme.

Kernel particle filter: iterative sampling for efficient visual tracking

Particle filter has recently received attention in computer vision applications due to attributes such as its ability to carry multiple hypotheses and its relaxation of the linearity assumption. Its shortcoming is increase in complexity with state dimension. We present kernel particle filter as a variation of particle filter with improved sampling efficiency and performance in visual tracking. Unlike existing methods that use stochastic or deterministic optimization procedures to find the modes in a likelihood function, we redistribute particles by invoking kernel-based representation of densities and introducing mean shift as an iterative mode-seeking procedure, in which particles move towards dominant modes while still maintaining as fair samples from the posterior. Experiments on face and limb tracking show that the algorithm is superior to conventional particle filter in handling weak dynamic models and occlusions with 60% fewer particles in 3-9 dimensional spaces.

Data Mining Nursing Care Plans of End-of-Life Patients: A Study to Improve Healthcare Decision Making

PURPOSEnTo reveal hidden patterns and knowledge present in nursing care information documented with standardized nursing terminologies on end-of-life (EOL) hospitalized patients.nnnMETHODn596 episodes of care that included pain as a problem on a patients care plan were examined using statistical and data mining tools. The data were extracted from the Hands-On Automated Nursing Data System database of nursing care plan episodes (n = 40,747) coded with NANDA-I, Nursing Outcomes Classification, and Nursing Intervention Classification (NNN) terminologies. System episode data (episode = care plans updated at every hand-off on a patient while staying on a hospital unit) had been previously gathered in eight units located in four different healthcare facilities (total episodes = 40,747; EOL episodes = 1,425) over 2 years and anonymized prior to this analyses.nnnRESULTSnResults show multiple discoveries, including EOL patients with hospital stays (<72 hr) are less likely (p < .005) to meet the pain relief goals compared with EOL patients with longer hospital stays.nnnCONCLUSIONSnu2002 The study demonstrates some major benefits of systematically integrating NNN into electronic health records.

Eye tracking using Markov models

We propose an eye detection and tracking method based on color and geometrical features of the human face using a monocular camera. In this method, a decision is made on whether the eyes are closed or not and, using a Markov chain framework to model temporal evolution, the subjects gaze is determined. The method can successfully track facial features even while the head assumes various poses, so long as the nostrils are visible to the camera. We compare our method with recently proposed techniques and results show that it provides more accurate tracking and robustness to variations in view of the face. A procedure for detecting tracking errors is employed to recover the loss of feature points in case of occlusion or very fast head movement. The method may be used in monitoring a drivers alertness and detecting drowsiness, and also in applications requiring non-contact human computer interaction.

Cyclic articulated human motion tracking by sequential ancestral simulation

Accurate tracking of cyclic human motion in video data helps in developing computer-aided applications such as gait analysis, visual surveillance, patient rehabilitation, etc. This paper presents a novel technique for tracking cyclic human motion based on decomposing complex cyclic motion into components and maintaining coupling between components. The decomposition reduces the dimensionality of the problem and enables a graphical modeling of the articulated human body. The coupling between components is modeled by their phase relationship and represented as directed edges in Bayesian networks and undirected edges in Markov random fields. Such coupling is maintained in tracking through ancestral simulation (AS) and Markov potentials in a sequential Monte Carlo tracking framework. We show that the approach handles severe self-occlusion and foreign body occlusion with improved accuracy and efficiency.

IIR Filter Banks and Wavelets

In digital filtering applications, it is known that Infinite-duration Impulse Response (IIR) filters [108], [372], [418], [26] provide a computational advantage over Finite-duration Impulse Response (FIR) filters that are designed to meet the same magnitude specifications. In the early work on multirate digital signal processing, this advantage of IIR filters did not carry over to their use in sampling rate conversions. The initial disadvantage arose from the fact that a conventional approach to IIR filter design and realization does not allow the exploitation of rate conversion in reducing the computational burden. However, a different approach, based on the use of a polyphase structure, provides a powerful framework that proves to be common to a low-complexity implementation of both IIR and FIR filters for rate conversion. In this chapter we examine the properties and design of IIR filter banks for one-dimensional and two-dimensional signals.

Density propagation for tracking initialization with multiple cues [human motion visual tracking]

The paper presents an automatic initialization procedure for visual tracking of human motion. Instead of relying merely on low-level image features to give a single estimate of the initial human posture, the system seeks to find a set of samples that carries multiple hypotheses of the pose. By accumulating different image cues in the first 3-15 consecutive frames and combining dynamic information regarding human motion, the system builds a human body model for the person to be tracked from a video sequence and produces a sample set as an estimate of the posterior distribution of the initial posture. The sample set provides a good starting point for tracking with sequential Monte Carlo methods.