Speech Data Rate Reduction Part II: Applicability of Sensitivity and Error Analysis

This paper deals with the application of modern estimation techniques to the problem of speech data rate reduction. It is desirable to adaptively identify and quantitize the parameters of the speech model. These paramaters cannot be identified and quantized exactly; the performance of the predictor is thereby degraded and this could prevent data reduction. In many cases it is desirable to emply a suboptimal predictor in order to simplify the algorithms, and predictor performance is again degraded. This paper develops sensitivity and error analysis as a potential method for determining quantitatively how speech data reduction system performance is degraded by imprecise parameter knowledge or suboptimal filtering. An intended use of the sensitivity and error analysis algorithms is to determine parameter identification and model structure requirements of configuration concepts for adaptive speech digitizers. First, sensitivity and error analysis algorithms are presented that form the basis for the remainder of the work. The algorithms are then used to determine how imprecise knowledge of vocal tract parameters degrades predictor performance in speech. Transversal filters have previously been proposed for this application. The sensitivity analysis algorithms are then used to determine when and by how much the transverse filter is suboptimal to the Kalman filter. In particular, the question of how effectively a higher order of all-pole model approximates a system with zeros is answered, as this question is of considerable importance in speech. Finally, the physical significance of the innovations process in speech data rate reduction is studied.     

Security applications of computer vision

In an age which bears witness to a proliferation of Closed Circuit Television (CCTV) cameras for security and surveillance monitoring, the use of image processing and computer vision techniques which were provided as top end bespoke solutions can now be realised using desktop PC processing. Commercial Video Motion Detection (VMD) and Intelligent Scene Monitoring (ISM) systems are becoming increasingly sophisticated, aided, in no small way, by a technology transfer from previously exclusively military research sectors. Image processing is traditionally concerned with pre-processing operations such as Fourier filtering, edge detection and morphological operations. Computer vision extends the image processing paradigm to include understanding of scene content, tracking and object classification. Examples of computer vision applications include Automatic Number Plate Recognition (ANPR), people and vehicle tracking, crowd analysis and model based vision. Often image processing and computer vision techniques are developed with highly specific applications in mind and the goal of a more global understanding computer vision system remains, at least for now, outside the bounds of present technology. This paper will review some of the most recent developments in computer vision and image processing for challenging outdoor perimeter security applications. It also describes the efforts of development teams to integrate some of these advanced ideas into coherent prototype development systems     

Identification of Aircraft Stability and Control Parameters Using Hierarchical State Estimation

Previous attempts to identify aircraft stability and control derivatives from flight test data, using three-degrees-of-freedom (3-DOF) longitudinal or lateral-directional perturbation equation-of-motion models, suffer from the disadvantage that the coupling between the longitudinal and lateral-directional dynamics has been ignored. In this paper, the identification of aircraft stability parameters is accomplished using a more accurate 6-DOF model which includes this coupling. Hierarchical system identification theory is used to reduce the computational effort involved. The 6-DOF system of equations is first decomposed into two 3-DOF subsystems, one for the longitudinal dynamics and the other for the lateral-directional dynamics. The two subsystem parameter identification processes are then coordinated in such a way that the overall system parameter identification problem is solved. Next, a six-subsystem decomposition is considered. Computational considerations and comparison with the unhierarchically structured problem are presented.     

On Sonar Signal Analysis

Systems which utilize acoustic energy to explore an undersea environment are called sonars. This introductory and tutorial paper presents a discussion of active sonar signal analysis concentrating upon sinusoidal, linear FM, and pseudorandom echolocation pulses. Many previously published results concerning sonar signals and their haracteristics are integrated, collected, and presented in a unified form such as to portray typical signal design considerations.     

Speech Data Rate Reduction Part I: Applicability of Modern Estimation Theory

Efficient coding of continuous speech signals for digital representation has attracted much interest in recent years. The underlying aim of efficient coding methods is to reduce the channel capacity required to represent a signal to meet a specific reconstruction fidelity criterion. To achieve this objective, modern speech data compression techniques rely on two very similar procedures. One procedure uses predictive deconvolution which subtracts from the current signal value that portion which can be predicted from its past and thus removes redundancy in the speech by removing sequential correlation. The signal thus requires fewer bits for equivalent quantization error. The second procedure involves identification of a complete mathematical model of the speech producing mechanism. This involves determination of the characteristics of the source that drives this transfer function. Data reduction is again achieved since the rate of change of the parameters of the speech model is much smaller than the rate of change of the speech waveform. This paper develops these data reduction procedures in terms of modern estimation theory, specifically a Kalman filter model, and illustrates the utility of this model as an analysis tool by means of an example based on a uniform tube which provides a qualitative assessment of the potential of the technique for application to real speech signals.     

Optimum Estimation of Bit Synchronization

In digital communication systems, optimum estimation and detection algorithms require that precise knowledge of the bit transition time be known to the receiver before bit-by-bit detection can be made. This paper presents the derivation of Bayes or maximum a posteriori estimation algorithms for optimum estimation of bit timing. Performance of the optimum system is evaluated and suboptimal realizations suggested.     

Analysis of Modeling and Bias Errors in Discrete-Time State Estimation

This paper concerns the effects of modeling and bias errors in discrete-time state estimation. The newly derived algorithms include the effect of correlation between plant and measurement noise in the system. The effects of nonzero mean noise terms and bias errors are considered. With plant or measurement matrix errors, divergence can occur. The local or linear sensitivity approach to error analysis, where the sensitivity is defined as a partial derivative with respect to a variable parameter taken about the modeled value, will not show this divergence due to neglect of higher order terms. Approximate algorithms are presented which circumvent the problem inherent in the local sensitivity approach. These make use of a "conditional bias" concept which views system error as a bias, conditioned on knowledge of the state estimates. It is shown that the actual error in optimum estimation is orthogonal to the residue error for suboptimum estimation where the residue error is defined as the difference between the actual estimation error and the optimum estimation error. Two examples, one concerning an integrated navigation system, demonstrate the theoretical results.     

Estimation Using Stochastic Feedback with Applications to Integrated Navigation Systems

This paper discusses an approach to linear estimation through use of a "control" fed back into the system to cancel out the effect of disturbances or error signals. Although this approach has very restricted application, it has found important usage in integrated navigation systems where one subsystem is an inertial measurement system. This approach is shown to be suboptimal and is compared with the optimal with respect to estimation accuracy and sensitivity to modeling errors. The feedback approach to estimation is shown to be similar to error estimation and correction in which the error states of the system are estimated and external correction applied. For discrete estimation using the feedback approach it is shown that error variance and Kalman gains for one-stage prediction should be used. Two examples are considered which compare the feedback approach to the optimum estimation approach. The system of the first example is quite simple, but provides simple analytical comparisons of the two estimation approaches. The second example system consists of a single-axis inertial guidance system and an independent position measuring system. Accuracy and sensitivity to modeling errors are compared. Other advantages and disadvantages of the two estimation approaches are discussed.     

AMETHYST: automatic alarm assessment becoming a reality

The aim of the AMETHYST (AutoMatic Event auTHentication SYSTems) project is to encourage the development of a high performance perimeter detection system which combines Video Motion Detection (VMD) technology with another type of Perimeter Intrusion Detection System (PIDS). AMETHYST will automatically assess the cause of PIDS alarms and pass to an operator those alarms likely to be caused by an intruder. It will filter out alarms not likely to have a human cause. A previous paper explaining and exploring the AMETHYST concept was presented at the 1995 Carnahan Conference. Since then PSDB has produced a single channel AMETHYST demonstrator and placed a contract for the development of an eight channel prototype AMETHYST system. This updated paper gives details of the hardware and software used with these two systems. Also described is PSDB's approach to the development of AMETHYST's automatic assessment algorithms. These will combine current expertise from Video Motion Detection (VMD) and Intelligent Scene Monitoring (ISM) systems with the unique AMETHYST approach. AMETHYST analyses picture sequences from before and after an alarm instead of continuously analysing live video. Sequences are provided by a Loop Framestore, either connected to or part of the AMETHYST system. The algorithms will be assessed and developed using PSDB's growing collection of over 150 alarm sequences