Electrical and Computer Engineering Commons^™

Visual Speech Recognition Using Multiple Deformable Lip Models, Devi Chandramohan

Electrical & Computer Engineering Theses & Dissertations

Motivated by the fact that human speech perception is a bimodal process (auditory and visual), several researchers have designed and implemented automatic speech recognition (ASR) systems consisting of both audio and visual subsystems, and shown improved performance relative to traditional purely auditory systems. Several visual speech reading approaches have used deformable templates to model the shape of a speaker's lips. Deformable templates are models of image objects, which can be deformed by adjusting a set of parameters to match the object in some optimal way, as defined by a cost function. Using a single deformable lip model has disadvantages such …

The Role Of Frame Selection And Bispectrum Phase Reconstruction For Speckle Imaging Through Atmospheric Turbulence, Elizabeth A. Harpold

Theses and Dissertations

Frame selection using quality sharpness metrics have been shown in previous AFIT theses, to be effective in improving the final product of images obtained using adaptive optics. This thesis extends this idea to noncompensated speckle image data. Speckle image reconstruction is simulated with and without frame selection. Speckle images require the processing of hundreds of data frames. Frame selection is a method of reducing the amount of data required to reconstruct the image. A collection of short exposure image data frames of a single object are sorted based on sharpness metrics. Only the highest quality frames are retained and processed …

Analysis Of Ladar Range Resolution Enhancement By Sinusoidal Phase Modulation, Leonard T. Masters, Martin B. Mark, Bradley D. Duncan

Electro-Optics and Photonics Faculty Publications

The ability of a ladar system to resolve two or more separate returns from a combined echo is related to the effective correlation bandwidth of the pulse emitted by the ladar system. Phase modulation of an outgoing pulse introduces additional frequency components, which increases the effective correlation bandwidth of the pulse and thus improves the range resolution of the system. In this paper, we discuss the general theoretical basis for achieving improved range resolution using a modulated waveform and a matched filter receiver. We then demonstrate these concepts by considering the particular case of improved range resolution for a sinusoidally …

Heterodyne Ladar System Efficiency Enhancement Using Single-Mode Optical Fiber Mixers, Donald K. Jacob, Martin B. Mark, Bradley D. Duncan

Electro-Optics and Photonics Faculty Publications

A theoretical performance analysis of a heterodyne ladar system incorporating a single-mode fiber receiver has been performed. For our purposes, the performance parameters of interest are the coupling and mixing efficiency of the ladar receiver, as they relate to the overall system carrier-to-noise ratio. For a receiver incorporating a single-mode fiber mixer, the received and local-oscillator fields are matched both spatially and temporally at the detector, yielding 100% mixing efficiency. We have therefore focused our efforts on determining an expression for the efficiency with which a diffuse return from a purely speckle target can be coupled into the receiving leg …

Integrating Reconstructed Scatterometer And Advanced Very High Resolution Radiometer Data For Tropical Forest Inventory, David G. Long, Perry J. Hardin

Faculty Publications

A scientific effort is currently underway to assess tropical forest degradation and its potential impact on Earth's climate. Because of the large continental regions involved, Advanced Very High Resolution Radiometer (AVHRR) imagery and its derivative vegetation index products with resolutions between 1 and 12 km are typically used to inventory the Earth's equatorial vegetation. Archival AVHRR imagery is also used to obtain a temporal baseline of historical forest extent. Recently however, 50-km Seasat-A Scatterometer (SASS) Ku-band imagery (acquired in 1978) has been reconstructed to = 4-km resolution, making it a supplement to AVHRR imagery for historical vegetation assessment. In order …

Perceptual Based Image Fusion With Applications To Hyperspectral Image Data, Terry A. Wilson

Theses and Dissertations

Development of new imaging sensors has created a need for image processing techniques that can fuse images from different sensors or multiple images produced by the same sensor. The methods presented here focus on combining image data from the Airborne Visual and Infrared Imaging Spectrometer (AVIRIS) hyperspectral sensor into a single or smaller subset of images while maintaining the visual information necessary for human analysis. Three hierarchical multi-resolution image fusion techniques are implemented and tested using the AVIRIS image data and test images that contain various levels of correlated or uncorrelated noise. Two of the algorithms are published fusion methods …

Color Image Segmentation, Kimberley A. Mccrae

Theses and Dissertations

The most difficult stage of automated target recognition ATR is segmentation. Current AFIT segmentation problems include faces and tactical targets previous efforts to segment these objects have used intensity and motion cues. This thesis develops a color preprocessing scheme to be used with the other segmentation techniques. A neural network is trained to identify the color of a desired object, eliminating all but that color from the scene. Gabor correlations and 2D wavelet transformations will be performed on stationary images and 3D wavelet transforms on multispectral data will incorporate color and motion detection into the machine visual system. The thesis …

Simplification Of The Generalized Adaptive Neural Filter And Comparative Studies With Other Nonlinear Filters, Henry Steven Hanek

Theses

Recently, a new class of adaptive filters called Generalized Adaptive Neural Filters (GANFs) has emerged. They share many characteristics in common with stack filters, include all stack filters as a subset. The GANFs allow a very efficient hardware implementation once they are trained. However, there are some problems associated with GANFs. Three of these arc slow training speeds and the difficulty in choosing a filter structure and neural operator.

This thesis begins with a tutorial on filtering and traces the GANF development up through its origin -- the stack filter. After the GANF is covered in reasonable depth, its use …

Feature-Based Correlation Filters For Object Recognition, Samuel Peter Kozaitis, Wesley E. Foor

Electrical Engineering and Computer Science Faculty Publications

Using an optical correlator, we experimentally evaluated a binary phase-only filter (BPOF) designed to recognize objects not in the training set used to design the filter. Such a filter is essential for recognizing objects from actual sensors. We used an approach that is as descriptive as a BPOF yet robust to object and background variations of an unknown or nonrepeatable type. We generated our filter by comparing the values of spatial frequencies of a training set. Our filter was easily calculated and offered potentially superior performance to other correlation filters.

Optical Image Analysis Using Fractal Techniques, Samuel Peter Kozaitis, Harold Gregory Andrews, Wesley E. Foor

Electrical Engineering and Computer Science Faculty Publications

Using an optical technique, we classified images of natural terrain based on their fractal dimension. We calculated the fractal dimension from an optically generated power spectrum obtained with a magneto-optic spatial light modulator (SLM). By using the fractal dimension to classify images of natural terrain, our post processing was simpler that when a ring-wedge detector was used.

A Direct Method For Surface Structure Recovering Based On Uoff, Ping Lin

Theses

The unified optical flow field (UOFF) theory which can be used for estimating motion and recovering surface structure was recently established in [9, 10]. The direct method developed in [2, 3, 4, 6, 7] does not need to explicitly solve the optical flow field and to find feature correspondence. Based on the UOFF, a direct method in space domain is developed to reconstruct the curved surface structure characterized by an Nth degree polynomial equation from a pair of stereo images. The initial work on this new method was reported in [8, 11].

In this study, I basically work on simulation …

A Radiosity Approach To Realistic Image Synthesis, Richard L. Remington

Theses and Dissertations

The Air Force Institute of Technology (AFIT) has not had the capability to generate photorealistic images. This thesis is a research effort to implement such a system at AFIT, and to apply knowledge from the particle transport community to this problem. Investigation into various techniques for generating photorealistic images led to the conclusion that the radiosity method would serve as the foundation for development. This is because of the increased accuracy of the diffuse reflections with the radiosity method and the fact that the images produced with radiosity axe view-independent. Two distinct tools to generate photorealistic images axe described. The …

The Effect Of An Adaptive Optical System's Spatio-Temporal Response On Imaging Performance, Patrick M. Harrington

Theses and Dissertations

Ideally, an adaptive optical control system would have instantaneous temporal response and infinite spatial bandwidth. In real systems, the response time of the adaptive optical control system is limited by the integration time of the wave front sensor, the computational time of the control algorithm, and the actuator response time. Additionally, finite inter actuator spacing limits the deformable mirror's ability to reproduce spatial frequencies having a period less than twice this spacing. Although analyses general enough to account for both the temporal and spatial characteristics of the adaptive optical system exist, they are complex and require detailed information regarding the …

Feature-Based Correlation Filters For Distortion Invariance, Samuel Peter Kozaitis, Robert Petrilak, Wesley E. Foor

Electrical Engineering and Computer Science Faculty Publications

In an optical correlator, binary phase-only filters (BPOFs) that recognize objects that vary in a nonrepeatable way are essential for recognizing objects from actual sensors. An approach is required that is as descriptive as a BPOF yet robust to object and background variations of an unknown or nonrepeatable type. We developed a BPOF that was more robust than a synthetic discriminant function (SDF) filter. This was done by creating a filter that retained the invariant features of a training set. By simulation, our feature-based filter offered a range of performance by setting a parameter to different values. As the value …

Position Recovery And Motion Analysis From Stereo Image Sequences, Yuxin Xie

Theses

Based on four stereo images, a new concept of unified optical flow field has been developed recently [3]. It is a generalization of the optical flow in stereo imagery. A set of fundamental equations is established to characterize the unified optical flow field. Another set of equations is then derived from which 3-D structure can be reconstructed and its motion can be analyzed by the use of the unified optical flow field quantities. In this thesis, we use some synthetic images of a rotated conical structure to study and test the theoretical results. The simulation and results are presented, thus …

Optical Image Segmentation Using Wavelet Correlation, Steven D. Pinski

Theses and Dissertations

This research introduces an optical method of segmenting potential targets using wavelet analysis. Implementation of an optical Harr wavelet is fulfilled using a magneto-optic spatial light modulator (MOSLM). Two methods of controlling wavelet dilation are explored: (1) spatial filtering of a ternary modulated MOSLM; (2) a single aperture positioned in front of a binary modulated MOSLM. Segmentation is performed through Vander Lugt correlation of a binarized image with a binarized optical wavelet. Three different image binarization methods are investigated for use in the correlation scheme: (1) average pixel value over an entire scene; (2) localized 4 x 4 pixel average …

Image Analysis Using Gabor Transforms: Correlated With Human Saccadic Movement Data To Identify The Human Visual Search Strategy, Ralph J. St John

Theses and Dissertations

This thesis investigated whether a relationship exists between the Gabor Correlation Coefficient (GCC) magnitudes of fixation points for a set of images viewed by six subjects and the human visual search strategy. A couple of different relationships were investigated. First, the data were analyzed to determine if the GCC magnitudes of human fixation points predicted the time ordered sequence of human fixation points during the performance of a visual search. Second, the data were analyzed to determine if there was a significant difference between the GCC magnitudes of the human fixation points and the GCC magnitudes of a random set …

Gabor Segmentation Of High Resolution Synthetic Aperture Radar Imagery, Michael A. Hazlett

Theses and Dissertations

This thesis investigates the use of Gabor filters and a radial basis function (RBF) network for segmentation of high resolution (1 foot by 1 foot) synthetic aperture radar (SAR) imagery. Processing involved correlation between the SAR imagery and Gabor functions. Two methods for selecting the optimal Gabor filters are presented. This research used complex Gabor functions and operated on single polarization HH complex data. Following the selection f the proper Gabor filters, correlation coefficients for each image were calculated and used as features for the RBF network. Provided are results demonstrating how Gabor processing and a RBF network provide image …

Image Segmentation Using Affine Wavelets, Steven E. Smiley

Theses and Dissertations

This thesis discusses the use of the multiresolution representation and Radial Basis Function (RBF) neural networks to segment both FLIR and SAR imagery. The multiresolution approximation coefficients are used as features into the RBF network which learns to distinguish between different cultural and natural regions or objects. The wavelets used are Mallat's spline wavelet and Daubechies' compactly supported wavelets. Additionally, this thesis provides an explanation of wavelets in a tutorial manner. It introduces wavelet theory and discusses two different approaches to generating the multiresolution or wavelet representation.

Face Recognition With The Karhunen-Loeve Transform, Pedro F. Suarez

Theses and Dissertations

The major goal of this research was to investigate machine recognition of faces. The approach taken to achieve this goal was to investigate the use of Karhunen-Loe've Transform (KLT) by implementing flexible and practical code. The KLT utilizes the eigenvectors of the covariance matrix as a basis set. Faces were projected onto the eigenvectors, called eigenfaces, and the resulting projection coefficients were used as features. Face recognition accuracies for the KLT coefficients were superior to Fourier based techniques. Additionally, this thesis demonstrated the image compression and reconstruction capabilities of the KLT. This theses also developed the use of the KLT …

The Design And Testing Of A Personal Computer Controlled Scanning Acoustic Microscope For Orthopaedic Surgical Applications, Hubert Heinrich Berndt

Theses

This thesis describes scanning acoustic microscopy and ultrasonic image processing to characterize and evaluate materials of known properties, as well as novel materials. This work describes the assembly and testing of a PC based device for nondestructive evaluation of bone materials for usage in orthopaedics research. It includes the identification of some hardware problems and their solution, and it also describes different image processing techniques that have been developed and applied in this work. In a growing number of disciplines it has been discovered that an acoustic microscope can provide new information, especially about the elastic properties of specimens and …

A Mouse-Driven Interface For Androx Ics-400, Chin-Tien Chen

Theses

A mouse-driven interface is extremely useful for software performing image processing tasks. This thesis describes an implementation of a mouse-driven interface for the Androx ICS-400 image processing board. The Androx board is housed in a Sun 4/330 running a UNIX-based operating system. The main goal of this interface is to provide a consistent and friendly user interface for image processing tasks and integrating the Sun window libraries with the Androx libraries. Various image processing tasks, such as gray scale and binary morphological operations, histogram, thresholding, filtering, spot, logical operations, arithmetic operations, zooming, scrolling, rank and image acquisition are implemented in …

Androx Implementation Of Nonlinear Filter, Keh-Chao Chu

Theses

Nonlinear filters, such as morphological operations, play an important role in image processing. Several software implementations of nonlinear filters are presented in this thesis. There may exist more than one method to implement a specific operation. The methods used in this thesis are chosen to be most suited for the machine used according to its structure. These most suited methods are also generally best for software implementations on other machines as well. The machine used here is the Androx image computing workstation.

A Parallelogram Four-Frame Model In 3-D Motion And Structure Recovery Using Unified Optical Flow Field Approach, Zhichun Lin

Theses

In this paper a new concept of the unified optical flow field (UOFF) for 3 D motion analysis from a stereo image sequence is implemented which is an extension of the temporal optical flow formulations developed by Horn and Schunck. A four frame model is established and a spatial optical flow is also introduced.

In order to alleviate the problem of the strict requirement of two identical sensors in common stereo imagery a new imaging system is presented in this paper that needs only one camera, two plane mirrors and two switches to generate a phase shifted binocular sequence of …

Radar Image Processing For The Afit Anechoic Chamber, Brian K. Sanders

Theses and Dissertations

The purpose of this study was to begin development of an Inverse Synthetic Aperture Radar imaging capability for the AFIT anechoic chamber. This began with an evaluation of the capabilities and limitations of the existing radar system and the chamber itself for this application. Then, after deciding on the image processing approach, software had to be written to collect the data necessary for image processing. This constituted the majority of this study, and resulted in a versatile, user-friendly program that automates the process of collecting data for high-resolution radar images. The program checks that the data to be collected will …

Fractal Geometry Segmentation Of High Resolution Polarimetric Synthetic Aperture Radar Data, Joseph L. Brickey

Theses and Dissertations

This thesis investigated the potential of fractal dimension estimation for segmenting high resolution polarimetric synthetic aperture radar. The data used during this research were collected with the Advanced Detection Technology Sensor (ADTS) developed by Massachusetts Institute of Technology Lincoln Laboratory with Defense Advanced Research Projects Agency funding. ADTS is a fully polarimetric calibrated 35 GHz SAR with one foot impulse response. A method of applying the correlation dimension algorithm developed by Grassberger and Procaccia for estimating the dimension of time series data was implemented to estimate the correlation dimension of polarimetric SAR data. A threshold sensitivity study was performed to …

A Micro-Based Machine Vision System, Hsueh-Chung Michael Feng

Theses

In this thesis, a low-cost general purpose machine vision system based on a personal computer of the PC-XT/AT family and an electronic camera controller based upon a Motorola 6811C11 single-chip microcontroller has been designed and implemented. This work is focused on image formation. The system supports a long distance camera-computer configuration. A picture editor, generic image manipulation software was designed and implemented. A single computer can supervise multiple cameras.

Non-Parametric Dominant Point Detection Algorithms, Kuo-Wei Huang

Theses

Dominant point detection is one of the most important preprocessing steps for landmark-based shape recognition and point-based motion estimation. Two new methods for detecting dominant points are presented. Both methods do not require any input parameter and the dominant points obtained by these methods remain relatively the same even when the object curve is scaled or rotated.

In the first method, for each boundary point, a support region is assigned to the point based on its local properties. Each point is then smoothed by a Gaussian filter with a width proportional to its determined support region. A significance measure for …

A Code Verification System For Ink Jet Printing, Daniel Liang Hsu

Theses

This paper describes system integration of a code verification system. The system utilizes a feature matching

Wind Streamline Ambiguity Removal Of Microwave Scatterometer Data, Hans W. Zaepfel

Electrical & Computer Engineering Theses & Dissertations

There is a class of satellite microwave scatterometers that return oceanographic information which includes wind vectors with directional ambiguity. The first part of the ambiguity removal process is to reduce the data to a directional ambiguity of 180°. This thesis shows that wind direction change contains the information required for streamline ambiguity removal and that syntactic pattern recognition techniques can be used to locate the areas of wind direction change. Spreading the information along the areas of wind direction change results in the removal of the streamline ambiguity. The algorithm is implemented and the results are presented showing that it …