The needfor robust real-time, and low-power feature extraction solutions calls for an integrative ap-proach that combines adaptive signal processing, machine learning, and micropower VLS
Speckle Field Statistics as Laser Beam Quality Criterion
The laser speckle field has been shown as a promising sensor modality in a variety of adaptive optics applications [96, 97] In laser communications, the speckle field produced by the received beam projected on a moving rough surface conveys information of its spa- tial distribution that can be used to adaptively improve the quality of the reception This information is extracted by observing various spectral components of the speckle field as received by a single photo detector [100].
In this chapter, I present an analog VLSI filter bank chip for speckle field spectral analysis In general, this technique is applicable for target in the loop (TIL) geometries,where the goal is to improve the quality of the transmitted beam, e.g to concentrate the beam intensity on a small spot on the target In the TIL scenario, coherent (laser) light is transmitted through a distorting medium (turbulent air) to an extended target where it is scattered and received by a detector, possibly at a different place from the transmitter of the laser light Applications of the speckle-based analysis approach to adaptive optics include laser communication technology, laser welding and cutting, and medical and military di- rected energy systems Experimental results characterizing the chip on speckle field data are reported in the next section.
SpeckSpect: A Mixed-Signal VLSI Speckle Spectrum Analyzer
Speckle analysis based on a PC (personal computer) requires a long sample time to achieve reliable spectrum property computation However, the characteristic turbulence distorting time constant is about a millisecond Moreover, the PC is a non-real-time system in its nature We present a real-time VLSI chip for speckle spectrum analysis The real-time computation is achieved through parallel hardware architecture Analog continuous-time circuits are used to provide continuous output for target-in-the-loop control.
The implemented 32-channel filter banks chip architecture is shown in figure 2.1, which can be configured in either parallel or cascade filter bank topology The circuit design and characterizations were reported in Chapters 2 and 3 For this application, the parallel filter bank architecture was adopted A 16-channel filter bank was programmed to have the center frequencies spaced oligarchically from 5K hz to 80Khz and a constant quality factor of 2 The two bandpass filters in a channel are programmed to have the same center frequency and quality factor The calibrated channel responses are shown in Figure 6.1.The speckle field experimental system architecture is shown in Figure 6.3 The output from the photo receiver is fed directly into the designed chip to provide continuous-time image quality metrics The output of each channel corresponds to the signal energy in a certain frequency band This frequency band is determined by the programming bits The outputs are then digitized and sent to a PC where they are used to calculate the speckle- metrics to control the active optic elements.
Figure 6.1: Measured filter bank channel response.
Speckle Measurement Experiment
ExperimentalSetip ee ee 96
The speckle field analysis setup is shown in Figure 6.2 Its corresponding functional block diagram is shown in Figure 6.3 In the experiment, we used a He — Ne laser beam with a 632nm wavelength and 10mW power The laser beam is cleaned with a micro-optic pinhole combination and collimated by a laser collimator, shown in Figure 6.2 The size of the collimated beam is controlled by a diaphragm.
A lens (L;) is mounted on a moveable stage, focusing the collimated laser beam in its focal plane, which is 30cm away from the lens The moveable stage allows the focal plane to be positioned within a range of 10cm with an accuracy of about one micron A metal disk with a rough surface is placed within this range A motor mounted at the axis of the disk rotates at the speed of 25 revolutions per second Since the axis of the rotating disk and the optical axis of the system differs by 6.5mm, it gives a relative speed of the rough surface at the optical axis of lm/s.
Figure 6.2: Speckle field analysis experimental setup.
A beam splitter, positioned in the optical axis, directs the back reflected speckles to- wards a photo detector The speckle field is sensed by a single photo diode in the photo detector feeding an electrical (voltage) signal into the analog VLSI chip The chip contains the filter bank, which performs continuous-time spectral analysis of the speckle field over
Figure 6.3: Speckle field analysis system diagram.
16 frequency channels Each channel measures low pass filtered energy of rectified band- pass filtered speckle-field input with individually programmable center frequency ranging from 100A 2z to 100kHz.
A PCB (printed circuit board) has been designed to host the chip and provide a interface with the PC The PC programs the chip, samples the spectrum output, and provides control for closed-loop applications.
Sampled Static Speckle Field
Speckle Spectrum Statistics 2.0 20 000000 e eee 99
When the motor and metal disk in speckle filed analysis system (Figure 6.3) rotates, the speckle field statistics is continuously computed by the speckle spectrum analysis chip. For the following experiment, 16 channels of the chip were programmed with center fre- quencies log-spaced from 5/Chz to 80Khz The band pass filters in all channels were programmed with the same quality factor of 2.
Figure 6.6 shows the output from four selected channels as a function of distance of the metal disk from the focal plane As we can see, for all the channels, a unique maximum is achieved when the lens is right at the focal plane position This property means that the output from all the channels can be used as an image quality metric.
The curve on the top of Figure 6.7 (a) shows the ratio of channel one and channel five outputs as a function of the distance of the metal disk to the focal plane Due to the mismatch in fabricated analog circuits, offset errors need to be taken into account.
Therefore, a second plot is shown at the bottom of Figure 6.7 (a), where offset is determined and compensated A zoom-in view of this offset compensated plot is shown in Figure 6.7(b) The beam size in this experiment is about 50mm Due to the diffraction limit, the smallest possible spot size is about 5 microns As we can see, the ratio from the two channels would also be used as an effective image quality metric.
Conclusions ee 100
In this chapter, the analog continuous time filter bank chip was characterized for the speckle field statistical analysis Our experimental results have confirmed the dependence of the speckle field energy on the degree of focus of the projected laser beam Both the filter bank channel outputs and their ratios can be used as effective laser beam quality criteria.
Figure 6.4: Statics Speckle from Camera 1 (a): Focused; (b): Non-focused.
Figure 6.5: Statics Speckle from Camera 2 (a): Focused; (b): Non-focused.
Distance from focal plane (am)
Figure 6.6: The signal of four channels depending on the position of the focus plane relative to the rough surface The center frequencies of the band pass filters in the four channel are programmed as follows: Channel 0 ~ 5kHz, Channel | — 10kHz, Channel 3 — 20kHz,Channel 5 — 30kHz.
Distance from focal plane (um)
Distance from focal plane (um)}
(b)Figure 6.7: Channel ratio as image quality metric (a): ratio of Channel 5 and Channel 1;(b): zoomed in version of ratio of Channel 5 and Channel 1.
The main contributions of this dissertation lay in the development of algorithms and design of robust feature extraction frontend for pattern recognition and adaptive systems, efficiently implemented in analog VLSI to achieve real-time and low-power operations. Each of these contributions are listed in the following subsections and their resulting pub- lications.
7.1.1 Three Decades Programmable OTA and Filter Design
While custom VLSI has the advantage of miniature size and real-time computation, a designed hardware usually has a fixed function To provide extra flexibility, the FPAA (field programmable analog array) has become a very popular approach At the circuit design level, the approach combines advantages of both analog and digital worlds: the continuous time analog filtering to achieve low power consumption, and the digital programmable OTA to provide a way of calibrating the filter parameters and circumventing a large amount of analog voltage biases.
The traditional programmable OTA and OTA-C filter designs share the limitations of poor dynamic range, small Œ„„ tuning range, complex filter parameters tuning circuit, and apt to process introduced variations to some extent This thesis proposed a wide linear dynamic range (2.4V,,), three decades digital programmable OT A [30] The filter parame- ters are conveniently programable through the designed OTAs and the selectable capacitors. The filter nonidealities are modelled by a GLM (generalized linear model) To achieve a desired filter characteristic, a nonlinear optimization procedure is developed to find the best programming bits compensating the process dependent errors Experimental results have shown that sufficient programming accuracies are achieved for the designed applications.
7.1.2 Auditory Perception Model for Robust Speech and Speaker Recog- nition
Many auditory models have been proposed in the literature for robust feature extrac- tion However, all these models are designed only from a pure algorithmic prospective.The models proposed in the literature are thus computational complex and not amenable to custom system-on-a-chip VLSI implementation The main contribution is systematic opti- mization of an auditory perception model from a hardware implementation perspective.Rather than attempting to model the exact physiological detail, we focused on the ab- stractions of models for neural information processing that are simple and yet effective in capturing the essentials of the signal processing in the ear The proposed model abstracts the functionality of the human auditory by a massive parallel non-linear filter banks The proposed system is mainly composed of continuous-time filters, which are conveniently im- plemented as OTA-C filters The extracted robust speech features outperform the standardMFCC features for both speech and speaker recognition tasks with additive background noise of different statistics [62] Both tasks can be implemented by the same programmable hardware.
7.1.3 Real-Time VLSI Systems for Sonar and Speckle Field Signal
The designed three decades programmable (100H z to 100k Hz) feature extraction fron- tend also finds applications in sonar signal and speckle field signal processing The minia- ture, real time, low power nature of the designed VLSI frontend is perfect for the real-time adaptive and mobile system integration.
Biosonar features generated by the custom chip were subjected to SVM training and classification, achieving real-time computation and target recognition accuracy par to the software simulations [93] Speckle field statistical analysis experiments have also con- firmed the dependence of speckle field energy on the degree of focus of the projected laser beam Both the filter bank channel outputs and their ratios can be used as effective laser beam quality metrics [101].
The challenges to implementing a feature extraction system on a single chip lay in the design of programmable, miniature, yet low power computation unit The designed VLSI frontend offers both reconfigurable filter bank architectures and programmable filters The system is implemented in a single 3mm X 3mm VLSI chip containing 96 programmable filters and consuming 9mW power.
As a comparative study, some similar frontends reported in the literature are outlined here The implemented schemas include e A Biosonar system implemented using commercial discrete analog chips for sonar auditory feature extraction [88] e A DSP MFCC speech feature processor known as HP Smartbadge IV [19]. e A single chip speech spectrum analyzer implemented in switched capacitor (SC) circuits [102]. e Our designed robust speech, sonar, and speckle field time-frequency feature extrac- tion frontend using OTA-C circuits [62].
The comparative study with these frontend is summarized in Table 7.1 The design presented in this thesis offers the most functionality, smallest area and lowest power con- sumption.
Table 7.1: Comparative study of feature extraction systems System name hardware type area power Biosonar system Discrete ispPAC chips | 18cm x 20cm 6W
HP Smartbadge IV DSP, fixed-point a wearable badget | 1.7WSpeech spectrum analyzer VLSI, switch-cap 7.0mm x 6.5mm | 150mWRobust auditory system VLSI, OTA-C 3mm x 3mm 9mW
A K, Jain, R.P.W Duin, and Jianchang Mao, Statistical pattern recognition: a review, Pattern Anal- ysis and Machine Intelligence, IEEE Transactions on Volume 22, Issue 1, Jan 2000 Page(s):4 - 37 R.N Bracewell, The Fourier transform and its applications, McGraw-Hill, New York, 1986.
LT Jolliffe, Principal Component Analysis, Spring-Verlag, New York, 1986.
R.L Gorsuch, Factor analysis, L Erlbaum Associates Hillsdale, NJ, 1983.
K Fukunaga, W Koontz, Application of the Karhunen-Loeve expansion to feature selection and ordering, IEEE Trans Computers, 1970.
N Kumar, A.G Andreou, Heteroscedastic discriminant analysis and reduced rank HMMs for im- proved speech recognition, Speech Communication, 1998.
S Mika, B Schoelkopf, A.J Smola, K.R Mueller, Kernel PCA and De-Noising in Feature Spaces, NIPS, 1998.
R.L de Queiroz, DAF Florencio, 4 pyramidal coder using a nonlinearfilter bank, Acoustics, Speech, and Signal Processing, 1996.
L.R Rabiner, B Juang, A tutorial on hidden Markov models and selected applications in speech recognition, Proceedings of the IEEE, 1989.
B Scholkopf, A.J Smola, Learning with Kernels: Support Vector Machines, Regularization, Opti- mization, and Beyond, MIT press, Cambridge, MA, 2001.
C.A Mead, Analog VLSI and Neural Systems, Addison-Wesley, 1990.
S Chakrabartty, Design and Implementation of Ultra-Low Power Pattern and Sequence Decoders, Ph.D Thesis, Dept of Electrical and Computer Engineering, Johns Hopkins University, 2004.
G Strang and T Nguyen Wavelets and Filter Banks, Wellesley-Cambridge Press, Wellesley, MA,1996.
S Chang, Y Kwon, and Sung-Il Yang Speech feature extracted from adaptive wavelet for speech recognition Electronics Letters Volume 34, Issue 23, 12 Nov 1998 Page(s):2211 - 2213
J.N Gowdy and Z Tufekci, Mel-scaled discrete wavelet coefficients for speech recognition IEEE Int. Conf on Acoustics, Speech, and Signal Processing, 2000 Volume 3, 5-9 June 2000 Page(s):1351 -
R Bain, University of South Carolina, Music, Acoustic and Cognition course lecture note. http://www music.sc.edu/fs/bain/vc/musc726a/index.htm
M Gupta, and A Gilbert, Robust speech recognition using wavelet coefficient features [EEE work- shop on Automatic Speech Recognition and Understanding, 2001 2001 Dec, Page(s):445 - 448
Ww J Phillips, Wavelets and Filter Banks Course Notes, http://www.engmath.dal.ca/courses/engm6610/notes/notes.html, January 9, 2003.
B Delaney, N Jayant, M Hans, T Simunic, A Acquaviva, “A low-power, fixed-point, front-end fea- ture extraction for a distributed speech recognition system,” Proc IEEE Int Conf Acoustics Speech and Signal Processing, (ICASSP’2002), vol 1 pp 793-796, Orlando FL, May 13-17, 2002.
Y Kuraishi, K Nakayama, K Miyadera, T Okamura, “A single-chip 20-channel speech spectrum analyzer using a multiplexed switched capacitor filter bank,’ JEEE Journal of Solid-State Circuits, vol 19 (6), pp 964-970, Dec 1984.
W Liu, A.G Andreou, and M.H Goldstein, “Voiced-Speech Representation by an Analog Silicon Model of Auditory periphery,” /EEE 7 Neural Networks, 1992.
A Brambilla, G Espinosa, F Montecchi, E Sanchez, Noise optimization in operational transcon- ductance amplifier filters, IEEE International Symposium on Circuits and Systems, 1989.
K Bult and G Gleelen, “An Inherently Linear and Compact MOST-Only Current Division Tech- nique” IEEE Journal of Solid-State Circuits, vol 27 (12), Dec 1992.
H.A Alzaher, H.O Elwan, M Ismail, “A CMOS highly linear channel-select filter for 3G multi- standard integrated wireless receivers” [EEE Journal of Solid-State Circuits, vol 37 (1), Jan 2002.
S.M Park and C Toumazou, “A packaged low-noise high-speed regulated cascode transimpedance amplifier using 0.6 4 n-well CMOS technology” 26th European Solid-State Circuits Conference, Stockholm, Sweden, 19-21 Sept 2000.
S Mallya and J.H Nevin, “Design procedures for a fully differential folded-cascode CMOS opera- tional amplifier,’ JEEE Journal of Solid State Circuits, vol 24 (6), Dec 1989.
B Razavi,“Design of Analog CMOS Integrated Circuits,’ TATA McGraw-Hill 2001.
D Johns and K Martin,“Analog Integrated Circuit Design,” Wiley 1997.
M.D Chen, J Silva-Martinez, S Rokhsaz, M Robinson, “A 2-Vpp 80-200-MHz fourth-order continuous-time linear phase filter with automatic frequency tuning”, JEEE Journal of Solid-State Circuits, vol 38 , Oct 2003.
Y Deng, S Chakrabartty and G Cauwenberghs, “Three-Decade Programmable Fully Differential Linear OTA,” Proc IEEE Int Symp Circuits and Systems (ISCAS’2004), Vancouver Canada, May 23-26, 2004.
S Pavan and Y Tsividis, “Time scaled electrical networks-Properties and application in the design of programmable analog filters” in JEEE Trans Circuits Syst II, vol 47, pp 161-165, Feb 2000.
J.E Kardontchik, “Introduction to the design of transconductor-capacitor filters” in K/uwer Academic Publishers, Boston, 1992.
G.J Gomez, S.H.K Embabi, E Sanchez-Sinencio, M Lefebvre, “A nonlinear macromodel for CMOS OTAs” JEEE Int Sym on Circuits and Systems, 1995.
S Szczepanski, R Schaumann, “Nonlinearity Induced Distortion of the Transfer Function Shape in High Order OTA-C Filters”, Analog Integrated Circuits and Signal Processing, vol.3, pp.143-151, 1992
P McCullagh, J.A Nelder, P McCullagh, “ Generalized linear models”, Monographs on Statistics and Applied Probability, Chapman and Hall, 1989.
Z Malik, O Omeni, C Toumazou, and E Rodriguez- Villegas, “Response Surface Methods Applied to a Continuous-time Filter” to apprea /EEE Trans Circuits Syst, 2005
A Acero, “Acoustical and Environmental Robustness in Automatic Speech Recognition”, CMU Ph.D Thesis, 1993.
Y Gong, “Speech Recognition in Noisy Environments: A survey”, Speech Communication, vol 16, pp 261-291, 1995.
P.J Moreno, B Raj, and R.M Stern, “A Vector Taylor Series Approach For Environment-Independent Speech Recognition”, Proc of the Internation Conference ofAcoustic, Speech and Signal processing, Atlanta, GA, May 1996.
B Raj, R Singh, R.M Stern, "Inference of missing spectrographic features for robust speech recog- nition”, Proc International Conference on Spoken Language Processing, 1998.
Y Deng, X Li, C Kwan, B Raj, R Stern, “Continuous Feature Adaptation for Non-Native Speech Recognition”, submitted to Jnt Journal on Signal Processing, (IJSP’2006).
S.V Vaseghi, B.P Milner, “Noise-Adaptive Hidden Markov Models Based on Wiener Filters”, Proc. European Conf Speech Technology, Berlin, 1993, Vol II, pp 1023-1026.
C J Legetter and P C Woodland “Flexible speker adaptation using maximum likelihood lienar regression”, Proceesing ARPA Workshop on Spoken Language System Technology 1995
A P Varga, R Moore, J Bridle, K Ponting and M Russell “Noise compensation algorithms for use with hidden Markov model based speech recognition”, Proceedings of ICASSP, 1988.
M.J.F Gales, “Model-based Techniques for Noise Robust Speech Recognition” Ph.D Dissertation, University of Cambridge, 1995.
P.C Woodland, J.J Odell, V Valtchev, and S.J Young “The development of the 1994 HTK large vo- cabulary speech recognition system” In Proceedings ARPA Workshop on Spoken Language Systems Technology, pages 104-109, 1995.
P Lockwood and P Alexandre, “Root adaptive homomorphic deconvolution schemes for speech recognition in noise” Jn Proceedings ICASSP, volume 1, pages 441-444, 1994.
L Rabiner and B H Juang “Fundamentals of Speech Recognition” Prentice Hall, 1993.
Hal Daume III, “From Zero to Reproducing Kernel Hilbert Spaces in Twelve Pages or Less”, 2004.
L.E Baum, and G Sell, “Growth transformations for functions on manifolds”, Pacific J Math., vol.27, no.2, pp.21 1-227, 1968.
P.S Gopalakrishnan, et al, “An inequality for rational functions with applications to some statistical estimation problems”, JEEE Trans on Information Theory, vol 37, January 1991.
S Chakrabartty, Y Deng and G Cauwenberghs, “Robust Speech Feature Extraction by Growth Transformation in Reproducing Kernel Hilbert Space,” Proc IEEE Int Conf Acoustics Speech and Signal Processing (ICASSP’04), Montreal Canada, May 17-21, 2004.
H Hermansky “Perceptual linear predictive analysis of speech”, Journal of the Acoustical Society of America, 87:1738—1752, 1990.
S Dupont, J Luettin “Audio-visual speech modeling for continuous speech recognition” /EEE Transactions on Multimedia, 2000
C Kwan, X Li, D Lao, Y Deng, etc “Voice Driven Applications in Non-stationary and ChoticEnvironment”, /nternational Journal on Signal Processing, 2005 (accepted).
K Kasper, H Reininger, and D Wolf “Exploiting the Potential of Auditory Preprocessing for Ro- bust Speech Recognition by Locally Recurrent Neural Networks” JEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP '97), Volume 2, p 1223, 1997
D.S Kim, S.Y Lee, and R.M Kil, “Auditory processing of speech signals for robust speech recogni- tion in real-world noisy environments”, JEEE Trans Speech and Audio Processing, 1999
J Tchorz and B Kollmeier, “A Model of Auditory Perception as Front End for Automatic Speech Recognition,” J Acoust Soc Am., vol 106, pp 2040-2050, 1999,
O Ghitza, “Auditory nerve representation as a basis for speech processing,” in Advances in Speech Signal Processing, ed by S Furui and M.M Sondhi, 1992 (Marcel Dekker, New York), Ch 15, pp. 453-485.
R.F Lyon and C.A Mead, “An Analog Electronic Cochlea,” IEEE Trans Acoust., Speech, Signal Processing, vol 36, pp 1119-1134, July 1988.
E Fragniere, A van Schaik and E.A Vittoz, “An Analog VLSI Model of Active Cochlea,” in Neu- romorphic Systems engineering: Neural Networks in Silicon, T.S Lande, Ed., Kluwer Academic, 1998.
Y Deng, S Chakrabartty, and G Cauwenberghs, “Analog Auditory Perception Model for Robust Speech Recognition,” Proc IEEE Int Joint Conf on Neural Network (IJCNN’2004), Budapest Hun- gary, July 2004.
E Zwicker, G Flottorp and S.S Stevens, “Critical Bandwidth in Loudness Summation,” J Acoust. Soc Am., vol 29, pp 548-557, 1957.
G Cauwenberghs, “Monaural Separation of Independent Acoustical Components,” Proc [EEE Int. Symp Circuits and Systems (ISCAS’99), Orlando FL, vol 5, pp 62-65, 1999
D.A Reynoids, “An overview of automatic speaker recognition technology”, JCASSP IEEE Interna- tional Conference on Acoustics, Speech, and Signal Processing, 2002.
S Furui, “An overview of speaker recognition technology”, ESCA Workshop on Automatic Speaker Recognition, 1994.
G.R Doddington, M.A Przybycki, A.F Martin, etc, “The NIST speaker recognition evaluation- Overview, methodology, systems, results, perspective”, Speech Communication, 2000.
D.A Reynolds, T.F Quatieri, R.B Dunn, “Speaker Verification Using Adapted Gaussian MixtureModels”, Digital Signal Processing, 2000.
W.M Campbell, “A SVM/HMM system for speaker recognition”, /nt Conf on Acoustics, Speech, and Signal Processing, 2003.
R.J Mammone, X Zhang, R.P Ramachandran, “Robust speaker recognition: a feature-based ap- proach”, Signal Processing Magazine, IEEE, 1996.
F Beaufays, M Weintraub, “Model Transformation for Robust Speaker Recognition from Telephone Data”, [EEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP'97), Volume 2, p 1063, 1997.
B Xiang, U.V Chaudhari, J Navratil, G.N Ramaswamy, “Short-time Gaussianization for robust speaker verification”, International Conference on Acoustics, Speech, and Signal Processing, 2002.
D A Reynolds, R C Rose, “Robust text-independent speaker identification using Gaussian mix- ture speaker models”, Speech and Audio Processing, IEEE Transactions on, Volume 3, Issue 1, Page(s):72-83, Jan, 1995.
L Liu, J He, “On the use of orthogonal GMM in speaker recognition”, Acoustics, Speech, and Signal Processing, Volume 2, Page(s):845-848, 1999.
“YOHO Speaker Verification DataBase”, http://www.ldc.upenn.edu/Catalog/
P Atkins, “Tutorial introduction and historical overview of the need for heading sensors in sonar applications”, JEE Colloquium on Heading Sensors for Sonar and Marine Applications, Jan, 1994.
F Martinerie, S Brisson, “An overview of recent developments in target tracking, in the active air- borne sonar networks domain”, Systems, Man and Cybernetics, International Conference on Systems Engineering in the Service of Humans, Page(s):662-668 vol.3, Oct 1993
N Cristianini, J Shawe-Taylor, An Introduction to Support Vector Machines and other kernel-based learning methods, Cambridge, 2000.
J.C Evans, J.S Smith, J Lucas, “Improving navigation using active landmark recognition”, [EE Col- loquium on Autonomous Underwater Vehicles and their Systems - Recent Developments and Future Prospects, May 1996.
J Kim, R.A Pearce, N.M Amato, “Feature-based localization using scannable visibility sectors”, IEEE International Conference on Robotics and Automation, Page(s):2854 - 2859 vol 2, Sept 2003.
O Hinton, “Underwater acoustic telemetry for AUVs”, IEE Colloquium on Autonomous Underwater Vehicles and their Systems - Recent Developments and Future Prospects, May 1996.
H.D Griffiths, “Radar and sonar imaging-a tutorial introduction”, JEE Colloquium on Recent Devel- opments in Radar and Sonar Imaging Systems: What Next?, Dec 1995.
T.R Clem, “Sensor technologies for hunting buried sea mines”, Oceans '02 MTS/IEEE, Page(s):452-
D.N Lambert, D.J Walter, D.C Young, S.R Griffin, and K.C Benjamin, “Developments in acoustic sediment classification”, OCEANS '98 Conference Proceedings, Page(s):26 - 31 vol 1, 1998.
R.H Shizumura, W.W.L Au, G.C Moons, P.E Nachtigall, H.L Roitblat and R.C Hicks, “Real- Time Sonar Classification of Proud and Buried Targets Using Dolphinlike Echolocation Signals,” J. Acoustical Soc, Am., vol 102, p 3152-, 1997,
G.S Okimoto, R Shizumura, and D Lemonds, “Active biosonar systems based on multiscale signal representations and hierarchical neural networks,” in Detection and Remediation Technologies for Mines and Minelike Targets II, SPIE Proc., vol 3392, pp.316-323, 1998.
H.L Roitblat, W.W.L Au, PE Nachtigall, R.H Shizumura, and G.C Moons, “Sonar Recognition of Targets Embedded in Sediment,” Neural Networks, vol 8, pp 1263-1273, 1995.
R Timothy Edwards, “The BioSonar Front-end Signal Processor System Manual” on http://bach.ece.jhu.edu/ tim, Dec 2001.
R P Gorman and T J Sejnowski, “Learned classification of sonar targets using a massively parallel network”, JEEE Trans Acoust., Speech, Signal Process, 1998.
V.N Vapnik, Nature of Statistical Learning Theory, Springer-Verlag, 1999,
P Clarkson and P J Moreno “On the use of Support Vector Machines for Phonetic Classification”, IEEE International Conference on Acoustics, Speech, and Signal Processing, 1999.
G Cauwenberghs and T Poggio, “Incremental and Decremental Support Vector Machine Learning,” Ady Neural Information Processing Systems (NIPS*2000), Cambridge, MA: MIT Press, vol 13, 2001.
G Cauwenberghs, R T Edwards, Y Deng, R Genov, and D Lemonds, “Neuromorphic Processor for Real-Time Biosonar Object Detection,” JEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP’02), Orlando FL, May 13-17, 2002.
J.W Hardy, “Adaptive Optics for Astronomical Telescope,” Academic, New York, 1991.
S F Gull, “Developments in maximum-entropy data analysis”, Maximum Entropy and Bayesian Methods editor J Skilling, Kluwer Academic, Dordrecht, pages:53-71 ,1989.
V.V Anisimov, S.M Kozel, and G.P Lakshin, “Spectral properties of a random intensity field pro- duced by coherent radiation scattering on a moving rough surface”, Radiotek, Elektron, vol 15,pages:539-545, 1970.
L L Goldfischer, “Autocorrelation function and power spectral density of laser-produced speckle patterns”, J Opt Soc Am A vol 55, pages: 247-252, 1965.
R.A.Muller and A.Buffington, “Real-time correction of atmospherically degraded telescope images through image sharpening”, J Opt Soc Am, vol 64, pages: 1200-1210, 1974.
A.Buffington and F.S.Crawford and R.A.Muller and A.J Schwemin and R.G.Smits, “Correction of atmospheric distortion with an image sharpening telescope”, J Opt Soc Am, vol 67, pages: 298-303, 1977,
M.A Vortontsov, G W Carhart, D.V Pruidze, J.C Ricklin, D.G Voelz, “Image quality criteria for an adaptive imaging system based on statistical analysis of the speckle field”, J Opt Soc Am Vol.
Y Deng, G Cauwenberghs, E Polnau, G Carhart, and M Vorontsov, “Integrated Analog Filter bank for Adaptive Optics Speckle Field Statistical Analysis”, Optics & Photonics, SPIE, San Diego, CA, Aug 2005.
Y Kuraishi, K Nakayama, K Miyadera, T Okamura, ”A single-chip 20-channel speech spectrum analyzer using a multiplexed switched-capacitor filter bank,” JEEE Journal of Solid-State Circuits,vol 19 (6), Dec 1984 Page(s): 964 -970.