Any generic computer vision algorithm must be able to copewith the variations in appearance of objects due to differentillumination conditions. While these variations in the shadingof a surface may seem a nuisance, they in fact containinformation about the world. This thesis tries to provide anunderstanding what information can be extracted from theshading in a single image and how to achieve this. One of thechallenges lies in finding accurate models for the wide varietyof conditions that can occur. Frequency space representations are powerful tools foranalyzing shading theoretically. Surfaces act as low-passfilters on the illumination making the reflected lightband-limited. Hence, it can be represented by a finite numberof components in the Fourier domain, despite having arbitraryillumination. This thesis derives a basis for shading byrepresenting the illumination in spherical harmonics and theBRDF in a basis for isotropic reflectance. By analyzing thecontributing variance of this basis it is shown how to createfinite dimensional representations for any surface withisotropic reflectance. The finite representation is used to analytically derive aprincipal component analysis (PCA) basis of the set of imagesdue to the variations in the illumination and BRDF. The PCA isperformed model-based so that the variations in the images aredescribed by the variations in the illumination and the BRDF.This has a number of advantages. The PCA can be performed overa wide variety of conditions, more than would be practicallypossible if the images were captured or rendered. Also, thereis an explicit mapping between the principal components and theillumination and BRDF so that the PCA basis can be used as aphysical model. By combining a database of captured illumination and adatabase of captured BRDFs a general basis for shading iscreated. This basis is used to investigate materialclassification from a single image with known geometry butarbitrary unknown illumination. An image is classified byestimating the coecients in this basis and comparing them to adatabase. Experiments on synthetic data show that materialclassification from reflectance properties is hard. There aremis-classifications and the materials seem to cluster intogroups. The materials are grouped using a greedy algorithm.Experiments on real images show promising results. Keywords:computer vision, shading, illumination,reflectance, image irradiance, frequency space representations,spherical harmonics, analytic PCA, model-based PCA, materialclassification, illumination estimation

Cancer is a class of diseases in which a group of cells grow uncontrollably, destroy surrounding tissue and eventually spread to other parts of the body, often leading to death. According to the American Cancer Society cancer causes accounts for 13% of all deaths. Much of the time cancer can be treated if diagnosed early. Considerable study is currently being undertaken to investigate tissue properties and their use in detecting cancer at an early stage through non invasive and non surgical methods. Oblique Incidence Diffuse Reflectance Spectrometry (OIDRS) is one such method. This thesis reports the design, fabrication and testing of a new miniaturized optical sensor probe with "side viewing" capability for oblique incidence diffuse reflectance spectrometry. The sensor probe consists of a lithographically patterned polymer waveguides chip and three micromachined positioning substrates and source/collection fibers to achieve 45 degree light incidence and collection of spatially resolved diffuse reflectance. The probe was tested at the Mayo Clinic in Rochester Minnesota. The test results show that the probe is capable of collecting data which can be analyzed to select image features to differentiate the cancerous tissue from non cancerous tissue. Using these probes, diffuse reflectance of human esophageal surface has been successfully measured for differentiation of cancerous tissues from normal ones.

Many Americans are afflicted with inflammation of the colon. They are also at a higher risk of developing colon cancer. Confocal microscopy of bulk epithelial tissue has the potential to provide information on tissue structural properties that may be lost in the fixation and slicing procedures required for histopathology. Optical sectioning provides images in three dimensions capturing the organizational structure of cells and colon crypts throughout the entire colon. I have constructed a custom built fluorescence and reflectance confocal microscope for imaging molecular and morphological changes associated with development of inflammation in a mouse model. A confocal microscope is a point scanning system that removes out of focus light by placing a pinhole aperture in the conjugate image plane located in front of the detector. We have two sources, 488 nm and 811 nm, for fluorescence and reflectance imaging, respectively. A polygon scanning mirror and a galvanometer scanning mirror allow for a variable scan rate between 8 and 15 fps. The lateral resolution of the system is approximately 3 μm with an axial resolution of 6 μm and 4 μm for reflectance and fluorescence mode, respectively. As colon tissue becomes inflamed, there is a distinct change in the structure and architecture of the tissue. The colon crypts are no longer uniform in size or distribution throughout the tissue. Having a large field of view of 1mm2 allows for many colon crypts to be visualized within a single frame. Histology was performed on the same tissue imaged for the inflammatory study confirming the constructed confocal microscope’s ability to characterize inflamed tissue and the potential use for guided biopsy. Mosaicing, or image tiling, is an imaging technique that stitches single frames together to produce a much larger field of view. An extended frame with 1 mm x 2 cm field of view is achieved within seconds. This extended frame would allow mosaicing of the entire mouse colon much faster than conventional methods without loss of resolution. The acquired confocal images of colon tissue demonstrate the microscope’s ability to resolve cell nuclei lining the colon crypts within a relatively large field of view.

This dissertation describes the design, fabrication and testing of an oblique incidence diffuse reflectance spectrometry (OIDRS) system for in-vivo and noninvasive detection of epithelial cancer. Two probes were fabricated using micromachining technology, which plays a significant role in the probe development by enabling device miniaturization, low-cost fabrication and precise assembly. The fist probe was developed and clinically tested for skin cancer detection. This probe consists of three source fibers, two linear array of collection fibers and four micromachined positioning devices for accurate alignment of the fibers. The spatially resolved diffuse reflectance spectra from 167 pigmented and 78 non-pigmented skin abnormalities were measured and used to design a set of classifiers to separate them into benign or malignant ones. These classifiers perform with an overall classification rate of 91%. The absorption and reduced scattering coefficient spectra were estimated to link the anatomic and physiologic properties of the lesions with the optical diagnosis. The melanoma cases presented larger average absorption and reduced scattering spectra than the dysplastic and benign ones. A second probe was designed to demonstrate the feasibility of a miniaturized ?side viewing? optical sensor probe for OIDRS. The sensor probe consists of a lithographically patterned polymer waveguides chip and two micromachined positioning substrates. This miniaturize probe was used to measure twenty ex-vivo esophageal samples. Two statistical classifiers were designed to separate the esophageal cases. The first one distinguishes benign and low dysplastic from high dysplastic and cancerous lesions. The second classifier separates benign lesions from low dysplastic ones. Both classifiers generated a classification rate of 100%.

<p>Any generic computer vision algorithm must be able to copewith the variations in appearance of objects due to differentillumination conditions. While these variations in the shadingof a surface may seem a nuisance, they in fact containinformation about the world. This thesis tries to provide anunderstanding what information can be extracted from theshading in a single image and how to achieve this. One of thechallenges lies in finding accurate models for the wide varietyof conditions that can occur.</p><p>Frequency space representations are powerful tools foranalyzing shading theoretically. Surfaces act as low-passfilters on the illumination making the reflected lightband-limited. Hence, it can be represented by a finite numberof components in the Fourier domain, despite having arbitraryillumination. This thesis derives a basis for shading byrepresenting the illumination in spherical harmonics and theBRDF in a basis for isotropic reflectance. By analyzing thecontributing variance of this basis it is shown how to createfinite dimensional representations for any surface withisotropic reflectance.</p><p>The finite representation is used to analytically derive aprincipal component analysis (PCA) basis of the set of imagesdue to the variations in the illumination and BRDF. The PCA isperformed model-based so that the variations in the images aredescribed by the variations in the illumination and the BRDF.This has a number of advantages. The PCA can be performed overa wide variety of conditions, more than would be practicallypossible if the images were captured or rendered. Also, thereis an explicit mapping between the principal components and theillumination and BRDF so that the PCA basis can be used as aphysical model.</p><p>By combining a database of captured illumination and adatabase of captured BRDFs a general basis for shading iscreated. This basis is used to investigate materialclassification from a single image with known geometry butarbitrary unknown illumination. An image is classified byestimating the coecients in this basis and comparing them to adatabase. Experiments on synthetic data show that materialclassification from reflectance properties is hard. There aremis-classifications and the materials seem to cluster intogroups. The materials are grouped using a greedy algorithm.Experiments on real images show promising results.</p><p><b>Keywords:</b>computer vision, shading, illumination,reflectance, image irradiance, frequency space representations,spherical harmonics, analytic PCA, model-based PCA, materialclassification, illumination estimation</p>

Thesis (Ph. D.)--University of Texas at Austin, 2001. / Vita. Includes bibliographical references. Available also from UMI/Dissertation Abstracts International.

Thesis (M.S.)--Textile and Fiber Engineering, Georgia Institute of Technology, 2004. / Griffin, Anselm, Committee Member; Srinivasarao, Mohan, Committee Chair; Park, Jung O., Committee Member. Includes bibliographical references (leaves 101-105).

Title from PDF of title page (University of Missouri--Columbia, viewed on March 8, 2010). The entire thesis text is included in the research.pdf file; the official abstract appears in the short.pdf file; a non-technical public abstract appears in the public.pdf file. Dissertation advisor: Dr. Gang Yao. Vita. Includes bibliographical references.

Heart rate monitoring (HRM) is an essential tool for monitoring physical activity and as a diagnostic tool in the clinical setting. The ability to monitor heart rate gives users and clinicians vital information about the current condition of the cardiovascular system before, during, and after exercise. However, HRM requires a telemetric chest strap, and comfort, transmission and fit can become problems with the chest strap. New technology using photoplethysmography (PPG) has emerged recently to provide the possibility of HRM without a telemetric chest strap during exercise. The aim of this study was to determine if a new device could detect heart rate over a broad range of skin types (I-V), and whether what wavelength would be most suitable for detecting the signals. A light emitting diode (LED) based PPG system was used to determine heart rate by change in pulsatile blood flow on 22 apparently healthy individuals (11 male and 11 female, 20-59 years old) of varying skin type. Skin type was classified according to a questionnaire in combination with digital photographs with a skin type chart. Each subject was exposed to four different wavelengths (470 nm, 520 nm, 630 nm, and 880 nm) and multiple trials were conducted on each wavelength. Heart rate detection was represented by modulation of the incident light wave and normalized by saturation into a pulsatile waveform represented as modulation average. The 520nm wavelength classified as visible green light provided a significantly greater (p<0.001) ability to detect heart rate. Increasing levels of melanin, or darker skin type (Type V) showed decreased modulation however this trend was not significant (p<0.067). There was no significant interaction between the wavelength of light and the skin type. In conclusion, a PPG based device can detect heart rate across skin types and use of a green light wavelength provides an even greater resolution. / text

A polarization sensitive optical coherence tomography (PS-OCT) instrument is presented for the study of glaucoma. Glaucoma is the second leading cause of blindness worldwide and causes irreversible damage to the retina. This PS-OCT system was built to perform retinal imaging with a swept source laser providing a 28 kHz A-scan repetition rate. Thickness, phase retardation, birefringence and reflectance index measurements were taken from the primate eyes on a weekly or semi-weekly basis through the course of a 30 week study. Statistical analysis of these measurements indicates that the reflectance index is the earliest measured indicator of glaucomatous changes and a potential marker for early glaucoma diagnosis. / text