<p>Indoor navigation could be used in many applications to enhance performance in</p><p>its specific area. Anything from serious life critical tasks like aiding firefighters or</p><p>coordinating military attacks to more simple every day use like finding a desired</p><p>shop in a large supermarket could be considered. Smartphones of today introduce</p><p>an interesting platform with capabilities like existing, more clumsy, indoor</p><p>navigation systems. The iPhone 3GS is a powerful smartphone that lets the programmer</p><p>use its hardware in an efficient and easy way. The iPhone 3GS has a</p><p>3-axis accelerometer, a 3-axis magnetometer and hardware accelerated image rendering</p><p>which is used in this thesis to track the user on an indoor map. A particle</p><p>filter is used to track the position of the user. The implementation shows how</p><p>many particles the iPhone will be able to handle and update in real time without</p><p>lag in the application.</p>

<p>Face detection is a preprocessing step for face recognition algorithms. It is the localization of face/faces in an image or image sequence. Once the face(s) are localized, other computer vision algorithms such as face recognition, image compression, camera auto focusing etc are </p><p>applied. Because of the multiple usage areas, there are many research efforts in face processing. Face detection is a challenging computer vision problem because of lighting conditions, a high degree of variability in size, shape, background, color, etc. To build fully </p><p>automated systems, robust and efficient face detection algorithms are required. </p><p> </p><p>Numerous techniques have been developed to detect faces in a single image; in this project we have used a classification-based face detection method using Gabor filter features. We have designed five frequencies corresponding to eight orientations channels for extracting facial features from local images. The feature vector based on Gabor filter is used as the input of the face/non-face classifier, which is a Support Vector Machine (SVM) on a reduced feature </p><p>subspace extracted by using principal component analysis (PCA). </p><p> </p><p>Experimental results show promising performance especially on single face images where 78% accuracy is achieved with 0 false acceptances.</p>

As CMOS processes size continues to shrink, a number of factors limit the ability of analog circuit performance to scale with the process. These issues include smaller transistor intrinsic gains and lower supply voltages. However, scaling continues to increase the speed and decrease the power of digital circuits. In this thesis, an active time-based integrator is proposed to replace amplifiers. The integrator, implemented using highly digital ring oscillators, seeks to take advantage of benefits offered by technology scaling while negating the issues of low gain and low supply voltages. The proposed integrator topology is used in a 20MHz 4th order continuous-time analog filter. Designed in a 90nm CMOS process, the time-based continuous-time filter achieves superior noise and linearity performance compared to state-of-the-art conventional active RC filters in simulations. / Graduation date: 2012 / Access restricted to the OSU Community at author's request from June 17, 2011 - June 17, 2012

My research is concentrated on applications of linear algebra in engineering, including wavelet analysis and structured matrices and (currently) approximation of large dense matrices by a mosaic of low rank blocks. / Singapore-MIT Alliance (SMA)

Lookup tables are frequently used in many applications to store and retrieve keyvalue pairs. Designing efficient lookup tables can be challenging with constraints placed on storage, query response time and/or result accuracy. This thesis proposes Geometric filter, a lookup table with a space requirement close to the theoretical lower bound, efficient construction, fast querying speed, and guaranteed accuracy. Geometric filter consists of a sequence of hash tables, the sizes of which form a descending geometric series. Compared with its predecessor, Bloomier filter, its encoding runs two times faster, uses less memory, and it allows updates after encoding. We analyze the efficiency of the proposed lookup table in terms of its storage requirement and error bound, and run experiments on Web 1TB 5-gram dataset to evaluate its effectiveness.

A polarization-independent electro-optically tunable add/drop filter utilizing non-polarizing novel relaxed beam splitters has been developed in LiNbO3 at the 1.55μm wavelength regime. The operation of this filter is based on passive directional coupler type beam splitters and strain-induced phase-matched TE↔TM polarization mode converters on an asymmetric Mach-Zehnder interferometer waveguide configuration. Fabrication parameters for channel waveguides, relaxed beam splitters and polarization mode converters were optimized individually then integrated to produce the final device. Single mode channel waveguides for both TE and TM polarizations were realized by the diffusion of 7μm wide Ti strips into LiNbO3 substrate. Relaxed beam splitters were produced using Ti diffused waveguides in a directional coupler configuration with 3.5mm long coupling region, 0.6º bending angle, and separation gap of 11μm and 13μm between waveguides. Tunable TE↔TM polarization mode converters with 99.8%

Linear tuning the frequency of an erbium-doped fiber ring laser using both a Fabry-Perot filter and an electro-optic tunable filter has been experimentally demonstrated. The rate of frequency change is determined by monitoring the fringes produced by laser light transmitted through a fiber Fabry-Perot interferometer. The fiber ring laser has been tuned over a 50 nm spectral range when using the Fabry-Perot filter and a tuning rate of 16480 nm/s has been achieved. The spectral width of the laser is 0.049 nm and the nearest sidelobe to the main peak is more than 30 dB below the central lobe. When the electro-optic tunable filter is used, a spectral range of 11 nm is reached. The spectral width is 2.33 nm and is in close agreement with the filter theoretical results. The sidelobe to main peak difference is around 13 dB.

This work is concerned with the theoretical study and the implementation of algorithms for solving two particular types of nonlinear optimization problems, namely unconstrained and simple-bound constrained optimization problems. For unconstrained optimization, we develop a new algorithm which uses a filter technique and a trust-region method in order to enforce global convergence and to improve the efficiency of traditional approaches. We also analyze the effect of approximate first and second derivatives on the performance of the filter-trust-region algorithm. We next extend our algorithm to simple-bound constrained optimization problems by combining these ideas with a gradient-projection method. Numerical results follow the proposed methods and indicate that they are competitive with more classical trust-region algorithms.

ABSTRAT In this thesis, we focused on the fabrication and frequency-modulation of £f/4 mode dual-band solidly mounted resonators and filters. To accomplish the Bragg reflector, the RF/DC magnetron sputtering system with dual targets is adopted to deposit alternating layers of quarter-wavelength Mo and SiO2 thin films. We tune the fabrication parameters in accordance with the AFM measurement and achieve a low roughness of 2.9nm on a nine-layer reflector. The piezoelectric layer, aluminum nitride (AlN) thin films, is deposited on the Bragg reflector by means of reactive RF magnetron sputtering. We alter the distance between substrate and target is altered to deposit AlN with various c-axis tilting angle which results in longitudinal and shear acoustic waves at the same time. Furthermore, we use mass loading effect to modulate the resonance frequency and fabricate £k- ladder type filters. In various numbers of reflector layers, the optimum frequency response is obtained with return loss of -26dB, in a SMR on a 3.5 pair reflector. To investigate the relationship between longitudinal and shear resonance in different resonance frequency, we alter the c-axis tilting angle of AlN as well as various mass loading on the SMRs. Based on the experimental results, the ratio of longitudinal to shear resonance frequency remains a constant value despite various c-axis tilting angle of AlN and mass loading on the SMRs. In addition, the electromechanical coupling coefficient, Kt2, of shear resonance raises with the increase of c-axis tilting angle of AlN. Finally,we have successfully fabricated SMRs with frequency modulataion of 3,899.68 Hz-cm2/ng and £k-ladder type filters with 26 MHz bandwidth.

This thesis proposes a new bandpass filter prototype modified based on T-type coupled resonator architecture by considering the parasitic shunt capacitance effect. After derivation, the new prototype can be proved equivalent to third-order Chebyshev bandpass filter. It is easy to realize the new prototype circuit by utilizing the inductor and capacitor library established from electromagnetic simulations. The couplings between circuit components can cause some transmission zeros to enhance attenuation rate at stopbands. This thesis designs several bandpass filters embedded in 4-layer laminate package substrate with center frequency at 2.45GHz. The measurement results show that most of these filters can achieve less than 1.7dB insertion loss and more than 14dB return loss at passband, and more than 30dB attenuation at 950MHz, 4.8GHz and 7.2GHz. One of the filters has a size of 1.9¡Ñ2.7mm2, which is the smallest area for the currently reported bandpass filters embedded in the organic package substrate.