Allen M. Peterson

Transform domain LMS algorithm

The concept of transform domain adaptive filtering is introduced. In certain applications, filtering in the transform domain results in great improvements in convergence rate over the conventional time-domain adaptive filtering. The relationship between several existing frequency domain adaptive filtering algorithms is established. Applications of the discrete Fourier transform (DFT) and the discrete cosine transform (DCT) domain adaptive filtering algorithms in the areas of speech processing and adaptive line enhancers are discussed.

On the Computation of the Discrete Cosine Transform

An N -point discrete Fourier transform (DFT) algorithm can be used to evaluate a discrete cosine transform by a simple rearrangement of the input data. This method is about two times faster compared to the conventional method which uses a 2N -point DFT.

Scale and rotation invariant texture classification

The problem of classifying scaled and rotated texture images is addressed using a number of different approaches. The first approach extracts invariant features from texture images; moment invariant features and log-polar filter features are employed. The second approach follows a mental transformation procedure similar to the process of scaled and rotated shape recognition carried out by human beings. Texture images are rotated and scaled to a specific size and orientation which allows the application of a more general rotation-scale sensitive classification scheme. A two-stage estimation procedure is introduced to determine the required scaling and rotation factors. Simulations show that the mental transformation approaches outperformed the other approaches, giving a good averaged error rate of 10%.<<ETX>>

Adaptive vector quantization using a self-development neural network

A neural network model, called SPAN (space partition network), is presented. This model differs from most of the currently seen neural networks in that it allows a network to adapt its structure by adding neurons, killing neurons, and modifying the structural relationships between neurons in the network. An adaptive vector quantization source-coding system based on SPAN is proposed. The major advantage of using SPAN as the codebook of a vector quantizer is that SPAN can capture the local context of the source signal space and map onto a lattice structure. A fast codebook-searching method utilizing the local context of the lattice is proposed, and a coding scheme, called the path coding method, for eliminating the correlation buried in the source sequence is introduced. The performance of the proposed coder is compared to an LBG (Y. Linde, A. Buzo, and R.M. Gray, 1980) coder on synthesized Gauss-Markov sources. Simulation results show that, without using the path coding method, SPAN yields performance similar to an LBG coder; however, if the path coding method is used, SPAN displays a much better performance than the LBG for highly correlated signal sources. >

Energy considerations in multichip-module based multiprocessors

Multichip modules permit highly efficient implementation of tiled architectures. If the tiles are implemented in submicron CMOS, extremely highly computation rates can be achieved, but power dissipation becomes the principal factor limiting achievable levels of integration and performance. Some examples of tiled architectures are described. The feasibility and advantages of reduced voltage operation for reducing energy per operation in power constrained environments are discussed.<<ETX>>

Bistatic-Radar Detection of Lunar Scattering Centers with Lunar Orbiter I

Continuous-wave signals transmitted from Lunar Orbiter I have been received on Earth after they have been reflected from the surface of the moon. The frequency spectrum of the reflected signals is used to locate discrete, heterogeneous, scattering centers on the lunar surface. The scattering centers are probably distinguished from the surrounding terrain by a higher surface reflectivity. Continuous-wave bistatic radar could provide an important new method for the study and mapping of planetary surfaces.

FIR filtering by the modified Fermat number transform

Right-angle circular convolution (RCC) and the modified Fermat number transform (MFNT) are introduced. It is shown that a linear convolution of two N-point sequences can be obtained by a corresponding N-point RCC. It is also shown that the MFNT supports RCC so that a linear convolution can be computed by an N-point MFNT and its inverse plus N multiplies. >

Digital filter structures having low errors and simple hardware implementation

Sensitivity and roundoff errors can seriously limit the application of recursive digital filters in practice, particularly when the filters have poles near z = + 1 . A filter structure, based on digital incremental computers is proposed, which has low sensitivity, good error characteristics, and simple hardware implementation for pole locations close to z = + 1 . Expressions for the roundoff errors are derived and compared to those for conventional structures. A design procedure is suggested to implement the new filter structure given the transfer function. Simulation results are presented.

Bistatic-Radar Observation of Long-Period, Directional Ocean-Wave Spectra with Loran A

Bistatic-radar scattering from medium- to long-wavelength (80 to 200 meters) ocean waves has been observed with the use of loran A (1.85 megahertz) transmissions and a receiver located 280 kilometers away. The received echoes have been converted into a time-delay, Doppler-frequency map in which the effects of anisotropies in the ocean-wave spectra are clearly shown. The distribution of the echoes in delay-Doppler space is consistent with Bragg scattering from trains of dispersed ocean waves.

Multiple channel neural network model for texture classification and segmentation

A computational image analysis model that resembles the functioning of the brain is introduced. The multiple-channel neural network model consists of three stages: multiple-channel representation, neural network classification and spatial context correction. The model is implemented and applied to the problem of texture analysis. Gabor filters are involved to represent the textural patterns. Low misclassification rates are obtained. Composite textural images are also applied to the system and accurately segmented images are obtained. The usefulness of the model is demonstrated.<<ETX>>