Rate-Distortion Analysis of Pose Estimation via Multi-Aspect Scattering Data

A Hidden Markov Model (HMM) provides an efficient means of modeling multi- aspect scattering data, with each HMM state representing a contiguous set of target-sensor orientations over which the wave physics is approximately stationary. The HMM-estimated state sequence gives a good approximation of the target pose. Rate-distortion theory is used to develop an error bound for estimating the state sequence (pose), as a function of the number of codes employed by the discrete HMM. The rate is defined as the number of discrete-HMM codes used to quantize the data, and the distortion is the probability of error in estimating the state of each measurement in the sequence. The rate-distortion function is calculated via the Blahut algorithm, considering example scattering data from an underwater elastic target. The performance of a discrete HMM using Lloyd quantization is compared with the rate-distortion bound, and is found to be far from optimal. Bayes-VQ is then applied in the context of HMM-based pose estimation, demonstrating significant performance improvement when good prior information about the target is available. This context-based quantizer accounts for the Bayes risk of pose estimation.

[1]  Richard E. Blahut,et al.  Computation of channel capacity and rate-distortion functions , 1972, IEEE Trans. Inf. Theory.

[2]  M. A. Bush,et al.  Speaker-independent vowel classification using hidden Markov models and LVQ2 , 1990, International Conference on Acoustics, Speech, and Signal Processing.

[3]  Robert M. Gray,et al.  An Algorithm for Vector Quantizer Design , 1980, IEEE Trans. Commun..

[4]  Marco Gori,et al.  A survey of hybrid ANN/HMM models for automatic speech recognition , 2001, Neurocomputing.

[5]  Lawrence Carin,et al.  Markov modeling of transient scattering and its application in multi-aspect target classification , 2001, 2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221).

[6]  R. Gray,et al.  Combining Image Compression and Classification Using Vector Quantization , 1995, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  T. Cover,et al.  Rate Distortion Theory , 2001 .

[8]  Christoph Neukirchen,et al.  A continuous density interpretation of discrete HMM systems and MMI-neural networks , 2001, IEEE Trans. Speech Audio Process..

[9]  B. D. Guenther,et al.  Aided and automatic target recognition based upon sensory inputs from image forming systems , 1997 .

[10]  Lawrence Carin,et al.  Hidden Markov models for multiaspect target classification , 1999, IEEE Trans. Signal Process..

[11]  Robert M. Gray,et al.  Bayes risk weighted vector quantization with posterior estimation for image compression and classification , 1996, IEEE Trans. Image Process..

[12]  Gerhard Rigoll,et al.  Maximum mutual information neural networks for hybrid connectionist-HMM speech recognition systems , 1994, IEEE Trans. Speech Audio Process..

[13]  John Makhoul,et al.  Discriminant analysis and supervised vector quantization for continuous speech recognition , 1990, International Conference on Acoustics, Speech, and Signal Processing.

[14]  Lawrence Carin,et al.  Matching pursuits with a wave-based dictionary , 1997, IEEE Trans. Signal Process..

[15]  Thomas M. Cover,et al.  Elements of Information Theory , 2005 .

[16]  Lawrence R. Rabiner,et al.  A tutorial on hidden Markov models and selected applications in speech recognition , 1989, Proc. IEEE.

[17]  Salvatore D. Morgera,et al.  A Structural Look at Pattern Recognition from the Point of View of Rate-Distortion Theory , 1988 .

[18]  Lawrence Carin,et al.  Multiaspect Target Identification with Wave-Based Matched Pursuits and Continuous Hidden Markov Models , 1999, IEEE Trans. Pattern Anal. Mach. Intell..

[19]  John W. Fisher,et al.  Pose estimation in SAR using an information theoretic criterion , 1998, Defense, Security, and Sensing.

[20]  Jian Li,et al.  Efficient mixed-spectrum estimation with applications to target feature extraction , 1995, Conference Record of The Twenty-Ninth Asilomar Conference on Signals, Systems and Computers.

[21]  S. Richard F. Sims.  Data compression issues in automatic target recognition and the measuring of distortion , 1997 .

[22]  R. Gray,et al.  Applications of information theory to pattern recognition and the design of decision trees and trellises , 1988 .