Search Results (1 - 20 of 20 Results)

Sort By  
Sort Dir
 
Results per page  

ZHU, XIANGDONGWAVELET-BASED SIGNAL ANALYSIS FOR THE ENVIRONMENTAL HEALTH RESEARCH
MS, University of Cincinnati, 2004, Engineering : Mechanical Engineering
As a part of the Cincinnati Lead Program Project (CLPP), postural balance signals from postural sway motions of lead exposed children have been measured and studied in an effort to relate the motions and the blood-lead levels. We developed several signal analysis techniques including advanced time-frequency analysis such as the analytic wavelet transform and the wavelet based multi fractal analysis to the postural balance data that had been measured by the CLPP. Like many other types of biological signals, postural sway motion signals are highly transient and multi-fractal. The wavelet analysis is very well suited for analysis of such transient signals because it uses time-frequency atoms of different sizes that depend on the frequency to break down the signal. The Wavelet based Multiple-Fractal Formalism (WMFF) and the analytic wavelet transform are two techniques adopted in this research to study the postural sway motion characteristics of lead exposed children. The main goal pursued in this study is to develop quantitative metrics to relate the postural balance motion and the blood lead level of children. WMFF calculates singularity spectra of signals; therefore can be used to identify abnormalities of the signals. Theories and procedures of wavelet based multi-fractal analysis are studied. The global singularities and multifractalities are obtained for the postural sway signals of 13 low blood lead level and 10 high blood lead level children. Multi-fractal characteristics of the signals from the two groups are compared with each other by using various data representations. The result shows that the WMFF can be a very useful tool in studying the effect of lead exposure by characterizing the motions in quantitative metrics namely the maximum spectrum level and the spectrum. Various time-frequency signal analysis and representation techniques are also developed to aid qualitative analysis of the sway signals. The analytic wavelet transform technique is believed to be extremely useful because it has all the advantages of the wavelet analysis as well as those of the conventional Fourier transform. Possible applications of the analytic wavelet transform to medical signal analysis are suggested with some preliminary results.

Committee:

Dr. Jay Kim (Advisor)

Keywords:

Wavelet Transform; Analytic wavelet; Time-frequency analysis; Multifractal formalism; Global regularity; Lead exposure; Postural balance; Center of pressure

Renfrew, Mark E.A Comparison of Signal Processing and Classification Methods for Brain-Computer Interface
Master of Sciences (Engineering), Case Western Reserve University, 2009, EECS - Computer Engineering
Non-invasive Brain-Computer Interface (BCI) methods have been investigated for use in physical therapy of stroke patients with motor deficits. This study investigates several methods of feature extraction and classification for suitability for use in such therapy. Electroencephalographic (EEG) data were collected during a motor task from four healthy control subjects and three subjects with motor deficiencies resulting from stroke. The EEG data were filtered using autoregressive (AR), mu-matched, and wavelet decomposition (WD) methods. The filtered data were classified using Support Vector Machines (SVM) and a linear classifier. Wavelet filtering showed a statistically significant (p < 0:05) improvement in classification accuracy over AR filtering for one subject when using the linear classifier. SVMs showed a statistically significant improvement over the linear classifier for all filtering methods for three subjects. No difference in classification accuracy was seen between linear and nonlinear SVMs.

Committee:

M. Cenk Cavusoglu, PhD (Committee Chair); Janis Daly, PhD, MS (Committee Member); Wyatt Newman, PhD (Committee Member); Mark Dohring, PhD (Committee Member)

Subjects:

Biomedical Research; Electrical Engineering

Keywords:

EEG; electroencephalography; BCI; brain-computer interface; wavelets; discrete wavelet transform; autoregressive; support vector machines

Cheng, HuainingOrthogonal Moment-Based Human Shape Query and Action Recognition from 3D Point Cloud Patches
Doctor of Philosophy (PhD), Wright State University, 2015, Computer Science and Engineering PhD

With the recent proliferation of 3D sensors such as Light Detection and Ranging (LIDAR), it is essential to develop feature representation methods that can best characterize the point clouds produced by these devices. When these devices are employed in targeting and surveillance of human actions from both ground and aerial platforms, the corresponding point clouds of body shape often comprise low-resolution, disjoint, and irregular patches of points resulted from self-occlusions and viewing angle variations. The prevailing method of depth image analysis has the limitation of relying on 2D features that are not native representation of 3D spatial relationships. On the other hand, many existing 3D shape descriptors cannot work effectively with these degenerated point clouds because of their dependency on 360-degree dense and smooth point clouds.

In this research, a new degeneracy-tolerable, multi-scale 3D shape descriptor based on the discrete orthogonal Tchebichef moment, named Tchebichef moment shape descriptor (TMSD), is proposed as an alternative for single-view partial point cloud representation and characterization. It has the advantage of decomposing a complex 3D surface or volumetric distribution into orthogonal moments in a much compact subspace that is independent of learning datasets, thereby supports accurate, robust, and consistent shape search and pattern recognition in the embedded subspace. Complimentary to the proposed descriptor, a new voxelization and normalization scheme is proposed to achieve translation, scale, and resolution invariance, which may be less of a concern in the traditional full-body 3D shape analysis but are crucial requirements for discerning partial point clouds.

To evaluate the effectiveness of TMSD and voxelization algorithms for static pose shape search and dynamic action recognition, we built a first-of-its-kind multi-subject pose shape baseline consisting of simulated LIDAR captures of actions at different viewing angles. Compared to the other existing public datasets, our baseline has more subjects and viewing angle variations to support solid algorithm development and evaluation. Using the pose shape baseline, we developed single-view nearest neighbor (NN) search for pose shape retrieval using TMSD. We proved the lower bounding distance condition under the orthonormality of Tchebichef moment, which prevents false dismissal by any subspace queries. Our experimental results show that 3D TMSD performs significantly better than 3D Fourier transform (3D DFT) and slightly better than 3D wavelet transform (3D DWT). It is also more flexible than 3D DWT for multi-scale representation because it does not have the restriction of dyadic sampling. The action recognition was built on the Naive Bayes classifiers using temporal statistics of a 'bag of pose shapes'. Our experiments demonstrate that the 3D TMSD-based classification of action and viewing angle outperforms the similar classification based on the depth image analysis using the popular 2D features of the histograms of oriented gradients.

In other experiments, we demonstrated our approach's scale invariance by showing consistent query and classification performance across a wide range of spatial scales, down to the extremely small scale of 6% of the original point clouds, at which level the 2D depth image analysis tends to degrade significantly. We also validated performance against varying viewing angles on both azimuth and elevation directions, which has an important implication for aerial sensor platforms.

In summary, many of the performance advantages shown by TMSD are fundamentally due to its sound mathematical properties. Through the direct 3D encoding of point cloud distribution, our research offers a promising new approach for analyzing low-quality, single-view 3D sensor data, other than the usual approach of 2D-based depth image analysis.

Committee:

Soon Chung, Ph.D. (Advisor); Nikolaos Bourbakis, Ph.D. (Committee Member); Yong Pei, Ph.D. (Committee Member); Vincent Schmidt, Ph.D. (Committee Member)

Subjects:

Artificial Intelligence; Computer Engineering; Computer Science; Information Science

Keywords:

3D shape descriptor; Tchebichef moment; Fourier transform; Wavelet transform; Point cloud; LIDAR; 3D shape search; Action recognition

Koglin, Ryan WEfficient Image Processing Techniques for Enhanced Visualization of Brain Tumor Margins
Master of Science in Engineering, University of Akron, 2014, Biomedical Engineering
Each year approximately 8 million people die from cancer on a global scale. Treatment varies depending on the stage and type of cancer but frequently includes surgery, radiation, and chemotherapy. For surgical removal of cancer, it is critical that health care professionals only remove the cancerous portion of tissue and avoid damaging healthy tissue. Imaging modalities are frequently used during surgery but are currently limited in their ability to differentiate between healthy and cancerous tissue. Image processing has the potential to allow surgeons the ability to visualize these differences. This study is aimed to develop an image processing algorithm capable of differentiating between healthy and cancerous tissues from a brain tumor. Fluorescence imaging was utilized to capture grayscale images of a mouse brain tumor samples, marked with green fluorescent protein-labeled biomarker, approximately 10 micro-meters thick. The discrete wavelet transform was then applied in conjunction with a nonlinear mapping function to process the images. Multiple levels of the discrete wavelet transform were applied to further differentiate between the healthy and cancerous tissue. A threshold was then applied and contour maps are shown for clarity. The results indicate both a clear in contrast and a successful segmentation of the tumorous region in each of the input images. This is shown through the statistical texture analysis, a comparison to previous studies, and by visual inspection.

Committee:

George Giakos, Dr. (Advisor); Bing Yu, Dr. (Committee Member); Narender Reddy, Dr. (Committee Member)

Subjects:

Biomedical Engineering

Keywords:

Image Processing Margins DWT Discrete Wavelet Transform Contrast Enhancement Non-Linear Mapping Functions Multi-Resolution Analysis Cancer Delineation

Shen, Chia-HsuanAcoustic Based Condition Monitoring
Doctor of Philosophy, University of Akron, 2012, Mechanical Engineering
Acoustic/vibration signal has traditionally benefited the condition monitoring of machinery. It can be further implemented in other field of applications as long as patterns associated with the conditions can be established. The dissertation consists of two areas of study, namely the gearbox condition monitoring and heart sound based diagnosis. The gearbox in a helicopter is a critical component with little response time prior to the failure. Therefore constant monitoring is necessary to prevent catastrophes. The developments of indicative parameters for condition monitoring of the gearbox remains to be a research of interest. Approximately more than 90% of heart murmurs are diagnosed to be normal and can be effectively determined by cardiac auscultation alone. However, current cardiology practices have been heavily relying on the expensive imaging equipment. With ever increasing national medical cost, a more optimized use of the high tech equipment is necessary. From the different sources of the acoustics, the objectives of the present dissertation concerned the analysis, the development of feature/parameter extractions, and/or the development of a pattern classifier for condition monitoring. In the gearbox condition monitoring, the vibration signatures of different gear and bearing damage scenarios were used to develop potential indicative parameters to detect gearbox faults. In heart murmur diagnosis, the study introduced a modular approach to computer-aided auscultation (CAA), where an alternative murmur characterization based on their acoustic qualities could be used. The analysis, the numerical characterizations, and the classifications for different types of the acoustic quality of murmurs as well as the classifications of the innocent murmurs and the pathological murmurs were carried out. In each of the topics of interest, analysis was performed in the time domain, the frequency domain and the time-frequency domain to acquire insights into the nature of the acoustic patterns under different conditions. Techniques such as the Fourier Transform (FT), the Continuous Wavelet Transform (CWT), and the Wigner-Ville distribution (WVD) were used for analysis for different signal domains. Possible features were than extracted to classify different representing conditions. The types of parameters/features extracted include the FT based features, the CWT based features, the Discrete Wavelet Transform (DWT) based features, and the Singular Spectrum Analysis (SSA) based features. The suitable features were selected based on techniques such as the Receiver Operating Characteristic (ROC) curve analysis and the Sequential Floating Forward Selection (SFFS) algorithm. The pattern classifiers used in the present dissertation include the K-Nearest Neighbor (KNN) classifier and the Classification and Regression Trees (CART). In gearbox condition monitoring, time-frequency analysis based on the CWT was considered a better visual examination solution among other considered techniques. Parameters based on the frequency components associated with the operating conditions were developed for damage identifications. In heart murmur diagnosis, different heart murmur qualities were quantitatively characterized by four extracted parameters based on their frequency characters and signal structure features. The parameters were able to correlate with the hemodynamics and physiology of the heart. Using the ROC curve analysis and the KNN classifier, an overall average accuracy of 87% was achieved. By using Sequential Floating Forward Selection (SFFS) and CART classifier, the average classification performance of different murmur qualities of up to 92% could be achieved. 90% accuracy was achieved for innocent murmur classification.

Committee:

Fred K. Choy, Dr. (Advisor); Shengyong Wang, Dr. (Advisor); Francis Loth, Dr. (Committee Member); Xiaosheng Gao, Dr. (Committee Member); Dale Mugler, Dr. (Committee Member); Kevin Kreider, Dr. (Committee Member)

Subjects:

Mechanical Engineering

Keywords:

gearbox health monitoring; heart sound analysis; wavelet transform; signular spectrum analysis; psychoacoustics; pattern recognitionl; heart failure monitoring

Ramanathan, VenkatramParallelizing Applications With a Reduction Based Framework on Multi-Core Clusters
Master of Science, The Ohio State University, 2010, Computer Science and Engineering
Data mining has emerged as an important class of high performance applications. At the same time, most parallel platforms today are clusters of multi-core machines. Thus, one of the major challenges today is achieving programmability and performance for data mining applications on multi-core machines and cluster of multi-core machines. FREERIDE (FRamework for Rapid Implementation of Datamining Engines) is a middleware developed based on the observation that the processing structure of a large number of data mining algorithms involves generalized reductions. FREERIDE offers a high-level interface and implements both distributed memory and shared memory parallelization. In this thesis, theWavelet Transformation algorithm is considered and it is shown how it can be modeled as a generalized reduction structure. It is parallelized using the FREERIDE middleware. It is shown that the algorithm can be parallelized in a communication and storage efficient manner. By this method, a good parallel efficiency with a speedup of around 42 on 64 cores is achieved. The second algorithm considered is a challenging new data mining algorithm, information theoretic co-clustering. This algorithm is parallelized using FREERIDE middleware. It is shown that the main processing loops of row clustering and column clustering of the the Co-clustering algorithm essentially fit into a generalized reduction structure. A good parallel efficiency is achieved and a speedup of nearly 21 is reported on 32 cores.

Committee:

Gagan Agrawal (Advisor); Radu Teodorescu (Committee Member)

Subjects:

Computer Science

Keywords:

Parallel wavelet transform; parallel co-clustering; reduction based framework

Zhang, YiSpatially Non-Uniform Blur Analysis Based on Wavelet Transform
Master of Science (M.S.), University of Dayton, 2010, Electrical Engineering
Object motion causes spatially varying blur in an image. Partial blur typically carries useful information about the scene. This information is useful for consumer imaging as well as computer vision. However, spatially varying blur also deteriorates image quality. The goals of our research are finding out this information and making images better. In this thesis we introduce a novel method for solving this partial blur problem. We define a statistical model of a spatially-varying blur image and estimate the local point spread function (PSF) by using a set of methods including double wavelet transform and local autocorrelation. Experimental results demonstrate the effectiveness of the proposed algorithm

Committee:

Keigo Hirakawa (Committee Chair); Eric Balster (Committee Member); Vijayan Asari (Committee Member)

Subjects:

Electrical Engineering

Keywords:

spatially non-uniform; motion blur dense estimation; over-complete wavelet transform; motion segmentation

He, ChaoAdvanced wavelet application for video compression and video object tracking
Doctor of Philosophy, The Ohio State University, 2005, Electrical Engineering

Wavelet transform has become a very powerful tool for image/video compression and processing. In this dissertation we present our research in the area of three-dimensional (3-D) wavelet based video compression and Gabor wavelet based video object tracking, respectively.

The wavelet transform has been successfully used for image compression. Its success in image compression motivates a lot of researches in the wavelet transform based video compression. 3-D wavelet transform based video compression is a direction drawing many attentions. We investigates two topics in this area.

The first topic is optimal 3-D zerotree (ZTR) structure for the 3-D wavelet based video compression. Many researches have been done to use the 3-D zerotree and 3-D wavelet transform in video compression. However, the problem of how to build a 3-D zerotree was not studied carefully. In this research, we study the theory behind building a zerotree and propose a new 3-D zerotree structure which generates better coding performance.

The second topic is time and space efficiency of the 3-D wavelet transform based video compression for real time video applications. The 3D wavelet transform needs to save and process large 3-D data which is a bottle neck for real time applications. We investigate this issue and present a time and space efficient video codec utilizing integer wavelet transforms.

The wavelet transform can represent signal in a multi-resolution way so it is also used in object tracking. We presents an object tracking method for object-based video processing which uses a 2-D Gabor wavelet transform (GWT), a 2-D triangular mesh and a 2-D golden section algorithm. The feature points are stochastically selected based on the energy of their GWT coefficients. The global placement of the feature points is determined by a 2-D mesh whose feature is the area of the triangles formed by the feature points. In order to find the corresponding object in the next frame, the 2-D golden section algorithm is employed, and this can be shown to be the fastest algorithm to find the maximum of a unimodal function.

Committee:

Yuan F. Zheng (Advisor)

Keywords:

3D wavelet; 3D zerotree; time and space efficient wavelet transform; object tracking

Ortiz-Rosario, AlexisImproved Methodologies for the Simultanoeus Study of Two Motor Systems: Reticulospinal and Corticospinal Cooperation and Competition for Motor Control
Doctor of Philosophy, The Ohio State University, 2016, Biomedical Engineering
The aim of this dissertation is to study methodologies and approaches used to enhance the understanding of how the corticospinal and reticulospinal systems cooperate and compete to recruit muscles. These motor systems have not been studied in a combined experimental design, nor have they been evaluated under novel analytical methodologies. The goal of this dissertation is to disseminate preliminary discovery of how these two motor systems combine their behavior under stimulation and propose novel methodologies to improve future work in this field. The dissertation will begin with a short introductory chapter (Chapter 1) followed by a review on brain-computer interfaces (Chapter 2) as a benchmark for what this work could lead to. The following chapters will encompass experiments performed to better understand and improve methodologies for these two motor systems. The experiments can be generally subdivided in neuroscience and biomedical engineering approaches of understanding these systems. Chapters 3 and 4 will explore neurophysiological circuits these systems might work through and how these two systems cooperate and compete to recruit muscles of the upper limb. Chapters 5 and 6 will present novel methodologies developed to help study these systems. Chapter 5 presents a novel isolation methodology using the wavelet transform and a statistical thresholding approach. Chapter 6 presents a methodology using multiple signal classification (MUSIC) to identify similar firing frequency profiles from cells to better understand their roles in movement. Chapter 7, the final chapter, will present closing remarks on how these discoveries and methodologies will help the field of neuroscience and biomedical engineering. This chapter includes improvements to previous experiments and future questions that rose from such. The final remark is how future brain-computer interfaces could potentially take information from subcortical motor systems (such as the ponto-medullary reticular formation) into account for future applications.

Committee:

Adeli Hojjat (Advisor); John A. Buford (Advisor); Thomas J Hund (Committee Member)

Subjects:

Biomedical Engineering; Computer Science; Neurosciences

Keywords:

reticulospina system, corticospinal system, signal processing, multiple signal classification, wavelet transform, gaussian mixture model, stimulus trains, principal component analsys

Chiang, TonyDesign and Evaluation of a Discrete Wavelet Transform Based Multi-Signal Receiver
Master of Science in Engineering (MSEgr), Wright State University, 2006, Electrical Engineering
General purpose receivers of today are designed with a broad bandwidth so that the receiver can accept a wide range of signal frequencies. These receivers usually accept one signal along with any interference that is included. To increase the signal detection capabilities of the wideband receiver, a design for a receiver that can detect two signals is needed. One of the requirements for this receiver is that the second weak signal needs to be processed in a timely manner so that the receiver can recognize it. To remedy the problem, a module was developed using wavelet-based techniques to remove spurs from the incoming signals to allow easier detection. The main basis for this concentration on wavelets comes from the way wavelets break down signals into portions (called resolutions) that allow easier determination of detail importance. Utilizing the multi-resolution attributes of the discrete wavelet transform, a way to remove signal spurs is made possible. When removing the signal noise from the signal, the two signal dynamic range of the system is increased, as this module is applied to multiple receiver systems for comparison of performance. Implementation of this system was originally done in C as well as MATLAB, but later is being implemented in VHDL with simulations done for verification of functionality.

Committee:

Chien-In Henry Chen (Advisor)

Keywords:

discrete; wavelet; transform; receiver; design

Zhu, XiangdongDevelopment and Applications of Analytic Wavelet Transform Technique with Special Attention to Noise Risk Assessment of Impulsive Noises
PhD, University of Cincinnati, 2008, Engineering : Mechanical Engineering

The development of the Analytical Wavelet Transform (AWT) as a transient signal analysis tool and the development of new noise metrics for possible future noise guidelines are the two major contributions of this dissertation research.

The AWT is developed and validated as a tool for transient signal analysis. Underlying theories and basic properties of the AWT are discussed in comparison with a commonly used short-time Fourier transform (STFT) method. The AWT is set up specifically for applications to noise and vibration analysis and applied to characterize highly impulsive sound and vibration signals. A new concept, 1/3 octave time history is defined and applied to the risk assessment of impulsive noise induced hearing loss. AWT is also applied to assess the performance of hearing protectors, to calculate the reverberation time of a room, and to characterize vibration signals of hand-tools. Some new concepts are developed taking the advantage of the capability of the AWT, which are time-frequency (T-F) and time-averaged noise reduction (NR) and frequency weighted time history.

The AWT is applied as a main signal processing tool to develop new noise metrics to assess the risk of impulsive noise induced hearing loss. Noise-induced hearing loss (NIHL) is the most common job-related illness. Current noise guidelines recommend the ii allowed noise exposure based on the equal energy hypothesis (EEH). The EEH based approach is appropriate for steady-state noises but not for impulsive noises because the time-averaging effect significantly underestimates the exposure risk. Because the Aweighted sound pressure level (SPL), a single valued metric, is used in current noise guidelines, risks of noises of vastly different temporal or spectral characteristic cannot be distinguished if their SPL levels are the same. To identify new noise metrics, fourteen new metric designs are developed which reflect the T-F characteristics of the noise obtained by the AWT in uniquely different ways. Statistical correlations of the metrics with the NIHL observed in a chinchilla noise exposure test are used to identify the best metric. The study identifies a few promising new noise metrics, which may be used to develop an improved noise guideline.

Committee:

Jay Kim (Committee Chair); David Brown (Committee Member); Mark Schulz (Committee Member); Seongho Song (Committee Member); William Murphy (Committee Member)

Subjects:

Engineering

Keywords:

Analytic Wavelet Transform; NIHL; Impulsive noise

Zhang, YiBlur Image Processing
Doctor of Philosophy (Ph.D.), University of Dayton, 2015, Electrical Engineering
Image blur stems from camera sensor pixel recording light from multiple sources. There are three causes of blur: object motion, optical defocus and camera shake. We propose double discrete wavelet transform (DDWT) to simplify the motion object and optical defocus blur analysis. In particular, DDWT de-correlates the blur from unobserved sharp image and DDWT coefficients give intuitive representation of blur kernel. DDWT based blur detection, estimation and deblurring are proposed to handle object motion blur image corrupted by low/high noise and defocus blur image. For camera shake blur, we propose inertial measurement unit (IMU) based deblurring. IMU is a set of motion sensors can be used to record the camera motion trajectory–the source of camera shake blur. Proposed work solves image blind deblurring problem by incorporating existing blind deblurring algorithm with IMU measurement in a complementary manner, along with image-IMU synchronization, therefore can be generalized by adopting other blind deblurring.

Committee:

Keigo Hirakawa (Committee Chair); Vijayan Asari (Committee Member); Tarek Taha (Committee Member); John Malas (Committee Member)

Subjects:

Electrical Engineering

Keywords:

Image blind deblurring, Noisy and blurry image, Bayesian statistics, double discrete wavelet transform, deblur with IMU, Motion estimation, Spatially varying blur

Sharma, NareshArbitrarily Shaped Virtual-Object Based Video Compression
Master of Science, The Ohio State University, 2009, Electrical and Computer Engineering

With the advancements in multimedia technologies, effective video compression has become more and more important. The size of the video files is always increasing with the increasing camera resolutions and because of never ending demand for better quality video signals. On the other hand, there is a limit to the available storage space and transmission bandwidth. Therefore, it is important to have good quality videos at low bit-rates. However, the widely established compression standards suffer from the annoying blocking artifacts at very low bit-rates and therefore, are not suitable for video coding at low bit-rates. The reason for the blocking artifacts is the usage of block based discrete cosine transform in these established compression methods. One good solution to this problem is to use wavelet transform which can be directly applied to the whole image at once, and thus, do not suffer from blocking artifacts. In addition, object based compression is also gaining in popularity because of the flexibility it provides to the end-user, and because of its widely believed potential to deliver good quality videos at very low bit-rates.

Therefore, this dissertation develops a compression method that further explores the advantages of object based compression, and uses shape adaptive wavelet transform for the coding of the arbitrarily shaped virtual-object thereby avoiding any blocking artifacts.

An arbitrarily shaped virtual-object compression method is developed. Method extracts the changing portion of the video as a 3D arbitrarily shaped virtual object from the non-changing portion termed as background. Arbitrarily shaped virtual object is coded using 3D wavelet compression whereas stationary background is coded using 2D wavelet compression. Experimental results demonstrate that the newly developed method outperforms 3D wavelet compression and the rectangular virtual-object compression by achieving higher compression ratio at a higher PSNR.

Committee:

Prof. Yuan F. Zheng, PhD (Advisor); Prof. Ashok Krishnamurthy, PhD (Committee Member)

Subjects:

Electrical Engineering

Keywords:

Arbitrarily shaped virtual-object; background; shape adaptive wavelet transform; 3D wavelet compression; 2D wavelet compression; object mask; texture coding; shape coding

Ghosh Dastidar, SamanwoyModels of EEG data mining and classification in temporal lobe epilepsy: wavelet-chaos-neural network methodology and spiking neural networks
Doctor of Philosophy, The Ohio State University, 2007, Biomedical Engineering
A multi-paradigm approach integrating three novel computational paradigms: wavelet transforms, chaos theory, and artificial neural networks is developed for EEG-based epilepsy diagnosis and seizure detection. This research challenges the assumption that the EEG represents the dynamics of the entire brain as a unified system. It is postulated that the sub-bands yield more accurate information about constituent neuronal activities underlying the EEG. Consequently, certain changes in EEGs not evident in the original full-spectrum EEG may be amplified when each sub-band is analyzed separately. A novel wavelet-chaos methodology is presented for analysis of EEGs and delta, theta, alpha, beta, and gamma sub-bands of EEGs for detection of seizure and epilepsy. The methodology is applied to three different groups of EEGs: healthy subjects, epileptic subjects during a seizure-free interval (interictal), and epileptic subjects during a seizure (ictal). Two potential markers of abnormality quantifying the non-linear chaotic EEG dynamics are discovered: the correlation dimension and largest Lyapunov exponent. A novel wavelet-chaos-neural network methodology is developed for EEG classification. Along with the aforementioned two parameters, the standard deviation (quantifying the signal variance) is employed for EEG representation. It was discovered that a particular mixed-band feature space consisting of nine parameters and LMBPNN result in the highest classification accuracy (96.7%). To increase the robustness of classification, a novel principal component analysis-enhanced cosine radial basis function neural network classifier is developed. The rearrangement of the input space along the principal components of the data improves the classification accuracy of the cosine radial basis function neural network employed in the second stage significantly. The new classifier is as accurate as LMBPNN and is twice as robust. Next, biologically realistic artificial neural networks are developed to reach the next milestone in artificial intelligence. First, an efficient spiking neural network (SNN) model is presented using three training algorithms: SpikeProp, QuickProp, and RProp. Three measures of performance are investigated: number of convergence epochs, computational efficiency, and classification accuracy. Next, a new Multi-Spiking Neural Network (MuSpiNN) and supervised learning algorithm (Multi-SpikeProp) are developed. Finally, the models are applied to the epilepsy and seizure detection problems to achieve high classification accuracies.

Committee:

Hojjat Adeli (Advisor)

Keywords:

Temporal Lobe Epilepsy; Electroencephalogram (EEG); EEG Classification; Epilepsy Diagnosis; Seizure Detection; Wavelet Transform; Chaos Theory; Artificial Neural Networks; Spiking Neural Networks; Principal Component Analysis; Cosine Radial Basis Function

Ouyang, DingxinIntelligent Road Control System Using Advanced Image Processing Techniques
Master of Science in Electrical Engineering, University of Toledo, 2012, College of Engineering
Over the past few years, Support Vector Machine (SVM) has been widely used in data classification field and has already been proved as an optimal solution for both linear and nonlinear classification problems. Since the image segmentation can be considered as a type of classification, SVM can be designed as an efficient image segmentation tool. This thesis aims to develop a SVM based intelligent road transportation control system, which involved three modules: pavement inspection, vehicle tracking, and collision warning. In the pavement inspection part, the SVM is used to extract the pavement from the background in a given image. The Radon transform is then applied to the pure pavement image to classify the crack to a particular type. In the vehicle tracking part, SVM trained by Gabor and edge features is involved to segment the first frame of a given video, which captured by an in-car camera. Another Wavelet feature based SVM is utilized to tracking this specific vehicle. In the collision warning part, the Time to Collision (TTC) is calculated by the scale change method. By the comparison between the TTC and a predefined threshold value, the Forward Collision Warning (FCW) system is designed, which can inform the driver to push the brake to avoid crash. Although the traditional image processing methods can fulfill all the three tasks above, limited success has been accomplished due to the low accuracy of image segment result. The proposed SVM algorithm can be trained by the proper feature, such as RGB feature, Gabor feature, Wavelet feature, etc., which makes the system appear to be more effective and computationally more efficient.

Committee:

Ezzatollah Salari (Committee Chair); Junghwan Kim (Committee Member); Jackson Carvalho (Committee Member)

Subjects:

Electrical Engineering

Keywords:

Image Processing; Support Vector Machine; Gabor Filter; Wavelet Transform; Hough Transform

Jackson, Brian PatrickAutomated Complexity-Sensitive Image Fusion
Doctor of Philosophy (PhD), Wright State University, 2014, Computer Science and Engineering PhD
To construct a complete representation of a scene with environmental obstacles such as fog, smoke, darkness, or textural homogeneity, multisensor video streams captured in diferent modalities are considered. A computational method for automatically fusing multimodal image streams into a highly informative and unified stream is proposed. The method consists of the following steps: 1. Image registration is performed to align video frames in the visible band over time, adapting to the nonplanarity of the scene by automatically subdividing the image domain into regions approximating planar patches 2. Wavelet coefficients are computed for each of the input frames in each modality 3. Corresponding regions and points are compared using spatial and temporal information across various scales 4. Decision rules based on the results of multimodal image analysis are used to combine the wavelet coefficients from different modalities 5. The combined wavelet coefficients are inverted to produce an output frame containing useful information gathered from the available modalities Experiments show that the proposed system is capable of producing fused output containing the characteristics of color visible-spectrum imagery while adding information exclusive to infrared imagery, with attractive visual and informational properties.

Committee:

Arthur Goshtasby, Ph.D. (Advisor); Jack Jean, Ph.D. (Committee Member); Thomas Wischgoll, Ph.D. (Committee Member); Lang Hong, Ph.D. (Committee Member); Vincent Schmidt, Ph.D. (Committee Member)

Subjects:

Computer Engineering; Computer Science

Keywords:

image registration; image fusion; stationary wavelet transform; entropy; image quality measures; fusion quality measures; adaptive image registration; multimodal image fusion; multimodal image processing; spatiotemporal analysis

MANTRALA, RAVI KSqueak and Rattle Detection: A Comparative Experimental Data Analysis
MS, University of Cincinnati, 2008, Engineering : Mechanical Engineering
Squeak and rattle evaluation is a common problem faced by automotive OEM’s. With increased importance to driving comfort and quality perception, squeak and rattle detection and elimination in modern automotive systems has become much more important in recent years. Many techniques involving time-frequency analysis, acoustics, digital signal processing, etc. are used to understand and control this unpredictable and undesirable vibro-acoustic phenomenon. An extensive literature survey was performed to gain understanding of the current state of the art. Detailed progression from the Fourier transform to the modern wavelet transform is also documented. This thesis is an attempt to perform a comparative analysis on experimental acoustic data collected on four, completely trimmed vehicles. These cars were tested on the four-axis, hydraulic road simulator at the Non-linear Testing Facility at the University of Cincinnati, Structural Dynamics Research Lab (UC-SDRL). The data was then analyzed using time-frequency techniques such as short time Fourier transform (STFT) and also with the advanced, complex Morlet wavelet technique. Both energy and amplitude normalization was performed on the Morlet wavelet and effects studied on data. The objective of the analysis was to detect and localize (in time and frequency) squeaks and rattles in automobiles from digital data recorded on the microphones when the hydraulic actuators were exciting the vehicle with certain signals. Understanding the applicability of the Morlet wavelets to the study of random bursts of acoustic energy and to observe their effectiveness in comparison to the standard STFT was also an important part of the study. Analysis of the data resulted in the conclusion that Morlet wavelets might not offer great advantage over the standard STFT in the process of detection of squeaks and rattles for automotive systems.

Committee:

Dr. Randall Allemang (Advisor)

Subjects:

Engineering, Mechanical

Keywords:

Wavelet Transform; Morlet; Squeak and Rattle

Jose, AdarshGene Selection by 1-D Discrete Wavelet Transform for Classifying Cancer Samples Using DNA Microarray Date
Master of Science in Engineering, University of Akron, 2009, Biomedical Engineering

Selecting a set of highly discriminant genes for biological samples is an important task for designing highly efficient classifiers using DNA microarray data. The wavelet transform is a very common tool in signal processing applications, but its potential in the analysis of microarray gene expression data is yet to be explored fully.

In this thesis, a simple wavelet based feature selection method is presented that assigns scores to genes for differentiating samples between two classes. The term ‘gene expression signal’ is used to refer to the gene expression levels across a set of pre-grouped samples. The expression signal is decomposed using several levels of the wavelet transform. The scoring method is based on the observation that the third level 1-D wavelet approximation of a gene expression signal captures the differential expression levels of the gene between two classes. The genes with the highest scores are selected to form a feature set to be used for sample classification. The method was implemented using MATLAB®. Experiments based on three real microarray gene expression datasets were carried out to examine the efficiency of the method. The classification performance of the method was compared to two standard filter based methods: the t-test and BSS/WSS methods using the 3-Nearest Neighbor Classifier. The results show that the wavelet-based method performs at least as well as the sum of squares and the wavelet based method in classifying cancer samples.

The results demonstrate that 1-D wavelet analysis can be a useful tool for studying gene expression patterns across pre-grouped samples.

Committee:

Dale Mugler, PhD (Advisor); Zhong-Hui Duan, PhD (Advisor)

Subjects:

Biomedical Research

Keywords:

discrete wavelet transform; microarray data; cancer; gene selection; classification

Cheng, Kai-JenCompression of Hyperspectral Images
Doctor of Philosophy (PhD), Ohio University, 2013, Electrical Engineering & Computer Science (Engineering and Technology)
Hyperspectral images contain a wealth of spectral data, and occupy hundreds of megabytes, which makes the transmission to remote reception sites more challenging and difficult. Thus, compression schemes oriented to the task of remote transmission are becoming increasingly of interest in hyperspectral applications. In this dissertation, we develop a transform-based coding for high-dimensional hyperspectral images. We study Shapiro's EZW algorithm according to multiple modifications and the results show that the asymmetric transform and tree design have best performance in compression; in addition, the data dependent algorithm results in more compact outputs. We also present the performance of hybrid transforms, including the discrete wavelet transform and Karhunen-Loeve transform, and the new asymmetric spatial-spectral tree structure. The results also show that the hybrid transform results in optimal energy distribution in spatial and spectral dimensions; moreover, the long spatial-spectral tree makes compression more efficient. We propose a Binary Embedded Zerotrees Wavelet (BEZW) algorithm for hyperspectral images. The zerotree quantization strategy of the BEZW is designed for the hybrid transformed images and the dual tree structures are defined in order to predict the insignificant pixels. For lossy hyperspectral image compression, the suitable quality criteria have to consider spectral information and reflect spectral loss. In this research we list spectral distortion measurements, examined distortion on lossy compression, and compare their abilities to accurately characterize compression fidelity in end user applications, such as unsupervised classification of image pixels. Finally, we cover the lossy and lossless results of the BEZW algorithm on AVIRIS datasets and comparisons of the conventional transform-based coders and the best predictive coders in terms of the complexity and distortion criteria. The BEZW algorithm is competitive with the best predictive algorithms and also is an efficient computational method which is comparable to transform-based algorithms.

Committee:

Jeffrey Dill (Advisor); Chris Bartone (Committee Member); Bryan Riley (Committee Member); Jundong Liu (Committee Member); Martin Mohlenkamp (Committee Member); Sergio Lopez- Permouth (Committee Member)

Subjects:

Computer Science; Electrical Engineering

Keywords:

Transform Based Coder; Wavelet Transform; Hyperspectral Image; Embedded Zerotree Wavelet; Karhunen Loeve Transform;

Lee, Soon GieHybrid Damage Identification Based on Wavelet Transform and Finite Element Model Updating
Doctor of Philosophy, University of Akron, 2012, Civil Engineering

Structural health monitoring (SHM) has gained more attentions recently since nearly 140,000 of a total 600,000 highway bridges in the US are nearing 50 years of age and are approaching the end of their design life. Most in-service highway bridge structures are suspected to be undergoing deterioration processes induced by the physical and harsh environmental changes. Therefore, timely maintenance with a robust SHM system having capability of early detection of impending damage is required to prevent catastrophic events for the public safety with reduced expenses.

Vibrational modal properties may not be sufficient for detecting early damage in local regions of complex civil infrastructure. Moreover, most of current damage detection methods require reference data which are not always available. There have also been pressing needs for real-time monitoring to prevent sudden catastrophic disasters. This dissertation addresses current challenges and needs identified in existing vibration-based damage detection methods, focusing on wavelet-based reference-free real-time damage identification and subsequent finite element model updating for quantifying damage severity.

First, a damage detection method based on a wavelet entropy analysis has been embedded in wireless smart sensor nodes (Imote2) and tested with three-story shear building and a laboratory truss bridge structure. To realize the reference-free damage detection, a continuous relative wavelet entropy (CRWE)-based damage detectionmethod is also proposed and demonstrated with a laboratory truss bridge structure. Although the reference-free CRWE method can detect damage locations without reference data,computational times put limitations in its applications to a real-time SHM system. To make real-time monitoring feasible in SHM systems, a statistical referencefree real-time damage detection method has been developed based on the wavelet packet transformation and log likelihood ratios.

Second, finite element model updating has been conducted to quantify the level of damage at the identified damage locations. For an identification model, fracturemechanics based cracked beam element with local flexibility coefficients and rotational spring stiffness coefficients have been used. After experimental modal testing of the laboratory truss structure, modal properties are extracted by the output-only frequency domain decomposition method. Because of limited number of sensors, mode shapes of each panel of the structure are separately extracted and combined by the interface DOF-by-DOF decentralized modal identification method. Modal properties (i.e. mode shapes and natural frequencies) are used to quantify physical damage level in term of crack depth.

In summary, this dissertation proposed wavelet-based robust and viable real-time reference-free damage localization methods and conducted damage quantification by finite element model updating. The proposed method has been experimentally verified and evaluated using test-beds that include a three-story shear building structure and a laboratory truss bridge structure.

Committee:

GunJin Yun, Dr. (Advisor); Ernian Pan, Dr. (Committee Member); Joan Carletta, Dr. (Committee Member); Kevin Kreider, Dr. (Committee Member); David Roke, Dr. (Committee Member)

Subjects:

Civil Engineering

Keywords:

Structural Health Monitoring; Damage Identification; Wavelet Transform; Damage Quantification; Finite Element Model Updating