Manohar Das

A bivariate autoregressive technique for analysis and classification of planar shapes

A bivariate autoregressive model is introduced for the analysis and classification of closed planar shapes. The boundary coordinate sequence of a digitized binary image is sampled to produce a polygonal approximation to an objects shape. This circular sample sequence is then represented by a vector autoregressive difference equation which models the individual Cartesian coordinate sequences as well as coordinate interdependencies. Several classification features which are functions or transformations of the estimated coefficient matrices and the associated residual error covariance matrices are developed. These features are shown to be invariant to object transformations such as translation, rotation, and scaling. Laboratory experiments involving object sets representative of industrial shapes are presented. Superior classification results are demonstrated. >

Lossless compression of medical images using two-dimensional multiplicative autoregressive models

Presents new methods for lossless predictive coding of medical images using two dimensional multiplicative autoregressive models. Both single-resolution and multi-resolution schemes are presented. The performances of the proposed schemes are compared with those of four existing techniques. The experimental results clearly indicate that the proposed schemes achieve higher compression compared to the lossless image coding techniques considered.

Nonstationary autoregressive modeling of object contours

A spatially variant circular autoregressive (SVCAR) model is introduced for the analysis and classification of closed shape boundaries. The model represents a closed shape boundary sequence as the output of a nonstationary all-pole linear system (driven by white noise) whose coefficients spatial evolution can be expressed as a truncated function expansion. Features derived from the SVCAR model are shown to be invariant to shape scaling, rotation, and translation. A shape-matching algorithm is developed to optimally adjust the SVCAR model coefficients for changes in contour sequence starting point. Laboratory experiments involving object sets representative of industrial, military, and geographic shapes are presented. Superior classification results are demonstrated. >

An efficient technique for modeling and synthesis of automotive engine sounds

This paper presents a new method for modeling and synthesis of automotive engine sounds using a deterministic-stochastic signal decomposition approach. First, the deterministic component is extracted using a synchronous discrete Fourier transform method and this is subtracted out from the original signal. Next, the (residual) stochastic component is modeled (and synthesized) using a new multipulse excited time-series modeling technique. The effectiveness of the proposed methodology is demonstrated using recorded data sets of actual engine sounds. The results of both numerical and subjective assessment tests are presented.

Real time implementation of adaptive noise cancellation

In this paper, we discuss the real time implementation of adaptive noise cancellation based on an improved adaptive Wiener filter on Texas Instruments TMS320C6713 DSK. Its performance is compared with the Leepsilas adaptive Wiener filter. LabVIEW models are illustrated for adaptive noise cancellation using National Instruments TI DSP test integration toolkit and adaptive filters toolkit. These models are tested with noisy wavelet test data sets and speech/wave files. Furthermore, a model based design of adaptive noise cancellation based on LMS filter using Simulink is implemented on TI C6713. The profile statistics of the auto-code generated by the Real Time Workshop for the Simulink model of LMS filter is compared with the dasiaCpsila implementation of LMS filter on C6713 in terms of code length and computation time. The signal to noise ratio of the filtered signal using improved adaptive Wiener filter improves by 2.5 to 4 dB as compared to Leepsilas adaptive Wiener filter. The dasiaCpsila code implementation of LMS filter on C6713 takes computation time of 205 ms and code length space of 1024 bytes whereas auto-code generated by Simulink takes computation time of 38.95 ms and 4032 bytes for code length.

An automated technique for image noise identification using a simple pattern classification approach

A simple pattern classification based noise identification method is proposed in this paper. The key idea involves isolation of some representative noise samples, and extraction of their statistical features for noise type identification. The isolation of representative noise samples is achieved using simple image filters and noise identification is performed using a few statistical and/or histogram based features. The method seems to be capable of accurately classifying the types of noise studied.

New studies on adaptive predictive coding of images using multiplicative autoregressive models

The authors introduce two new one-dimensional multiplicative autoregressive (MAR) models for adaptive predictive coding of digitized images. The proposed scheme offers a number of advantages. These include easy implementability, a high signal-to-noise ratio at a moderate bit rate, and guaranteed stability of the predictive coder. Results of extensive experimental studies are presented.

A simple sleep stage identification technique for incorporation in inexpensive electronic sleep screening devices

This paper investigates pattern recognition techniques for identification of sleep stages based purely on respiratory signals. It focuses on computationally simplistic methods, which can be implemented on an inexpensive microprocessor in a low-cost and comfortable home-screening device for the detection of sleep-related disorders, such as obstructive sleep apnea. In spite of the fact that sleep stages are defined by measurements of electrical activity in the brain, there are quantifiable changes in the respiratory pattern which can be used to distinguish between sleep stages with a reasonable degree of accuracy. Multiple respiratory features were evaluated for their efficacy in classifying each 30 second epoch of a respiratory signal as Wake, Non-REM, or REM sleep. Both linear and naive-Bayes classifiers were comparatively tested on nasal and abdominal respiration signals collected from MIT-BIH Polysomnographic database, but optimal results were achieved using a naive-Bayes classifier. The findings of this study support the feasibility of respiratory-based sleep stage classification, which can be refined to a technique accurate enough for inexpensive sleep monitoring devices.

Statistical signal modeling techniques for automated recognition of water-borne microbial shapes

The purpose of this paper is to present some preliminary results related to the problem of automated detection and identification of water-borne microbiota (bacteria, algae, and protozoa). The topics addressed include acquisition and creation of a microbiota image database, enhancement using Wiener/nonlinear filters, statistical modeling of shape contours, and classification.

Near-lossless compression of medical images

This paper studies the characteristic properties of a specific class of near-lossless image compression schemes which consists of a lossless coder followed by a uniform scalar quantizer. Three specific instances of such schemes are investigated; namely, differential pulse code modulation, hierarchical interpolation, and two-dimensional space-varying multiplicative autoregressive coders. The compression gains attainable with such schemes are studied and results of experiments conducted on several medical images are presented.