With an increasing emphasis on personal privacy, security, and convenience, the security of identification system is an important issue nowadays. In this thesis, two intelligent identification systems, laser speckle image identification system and laser-based finger biometric system, are proposed to perform superior solutions for identification applications. In laser speckle image identification system, we investigated the characteristics of laser speckle as well as proposed an appropriate algorithm to establish this system. The proposed algorithm is a coarse-to-fine process which identifies laser speckle images systematically. In laser-based finger biometric system, a new biometric approach is described to proceed personal identification using a scanner with a low power laser scans across the surface of the finger and continuously recording the reflected intensity at a fixed position. Experimental results show that the recognition rates of the proposed system are both 100%.

The use of breath odours in medical diagnosis dates back to classical times, though in its modern form the technique is only a few decades old. There are several breath tests in common clinical use, though all of them involve administration of a known or labelled exogenous compound. More recently, over the last twenty years, interest has focussed on analysis of endogenous metabolites in breath, but despite a large number of published studies reporting a number of disease markers, there has been little or no impact on clinical practice. Nonetheless, breath analysis offers a number of potential advantages over current biochemical methods. One major advantage of breath analysis is its non-invasive nature, which has led to significant interest in its use at point-of care for monitoring chronic diseases such as diabetes and the chronic infections ubiquitous in cystic fibrosis. However, breath analysis classically involves the use of expensive laboratory based analytical equipment which requires extensively-trained personnel and which cannot readily be miniaturised. Systems based on simple gas sensors might offer a way of overcoming these limitations. In recent years, Cranfield University has developed an instrument called the single metal oxide sensor gas analyser (SMOS-GA, more commonly referred to as the “Breathotron”) as a proof of concept for sensor-based breath analysis. In this project the Breathotron has been used in conjunction with selected ion flow tube mass spectrometry (SIFT-MS) and thermal desorption gas chromatography mass spectrometry (TD-GC-MS) to determine the changes in the concentrations of volatile organic compounds (VOCs) in breath in a number of experimental situations which a relevant to the diagnostic monitoring of diabetes mellitus. Studies conducted on clinically healthy volunteers were: an oral glucose tolerance test (OGTT); a six minute treadmill walking test; and a bicycle ergometer test. Additionally Breathotron and analytical data were also obtained during a hypoglycaemic clamp study carried out on hypoglycaemia-unaware Type I diabetics. The principle breath volatiles determined analytically were: acetone, acetaldehyde, ammonia isoprene though data on a number of others was also available. In general, it proved difficult to establish any reproducible relationship between the concentration of any compound measured and blood glucose concentration any of the experimental interventions. It was notable, though, that statistically significant associations were observed occasionally in data from individual volunteers, but even these were not reproduced in those trials which involved repeated measurements. This remained true even where spirometry data were used to derive VOC clearance rates. This may explain previous reports from smaller studies of an association between glucose and breath acetone concentration. It seems probable that any experimentally-induced changes in breath VOC concentration or clearance were of much smaller magnitude than background variability and was consequently not detectable. These observations were mirrored in the sensor-derived results. Multivariate analysis across all trials where Breathotron data were obtained suggested clustering by individual volunteer rather than glycaemic status. This suggests that that there exists a “background” breath volatile composition, dependent perhaps on such factors as long-term diet, which is independent of our experimental intervention. The Breathotron was also used as a platform to assess the performance of three different types of mixed metal oxide sensor in vitro. Calibration curves were generated for acetone, ammonia and propanol covering the physiological range of concentrations and with a similar water content to breath. Close correlations were obtained between concentration and the amplitude of the sensor response. Sensor response reproducibility was also determined using acetone at a concentration of 10ppm with dry and humidified test gas. There were significant differences between sensor types in overall reproducibility and in response to humidity. These results suggest that had there been substantial changes in breath VOC composition as a result of our experimental interventions, any of the types of sensor used would have been capable of responding to them. In summary, these results do not support the efficacy of breath VOC analysis as a means of non-invasive diagnostic monitoring.

SIFT-MS is a relatively new trace gas analysis technique that has wide application. One particular attribute of the instrument is the ability to detect and quantify volatile organic compounds to the parts per trillion in real-time without the need for sample preparation. However the issue of maintaining accuracy at these low concentrations required attention as it was evident large or polar analytes were being lost by adsorption to the SIFT instrument’s inlet system. The purpose of this research was to evaluate the performance of a passivated inlet in lowering any adsorption in the inlet system compared to the current unpassivated inlet of the SIFT instrument. Volatile concentrations of vanillin (C₈H₈O₃ 152.15 g/mol), ammonia (NH₃ 17.03 g/mol), and hydrogen sulfide (H₂S 34.08 g/mol) were measured. The results determined the passivated inlet provided a significantly better inlet response to these compounds. Consequently improved passivated inlets were installed on current models of SIFT-MS VOICE200®, and also the research laboratory VOICE100™ instrument. Having established a more reliable sampling system for very low concentrations of analyte, attention was paid to SIFT-MS flavour analysis of two foods, cheese and chocolate. The volatile matrix of these foods is highly complex and the compounds of interest are typically difficult to measure. The key aroma compounds for analysis were based on reported literature and earlier SIFT-MS studies which provided a useful framework for the current food flavour research. A significant finding from the SIFT-MS examination of Parmesan cheese is that differences in the relative concentration of some characteristic aroma compounds were a consequence of the milk type used in manufacture. Endogenous enzymes responsible for a multitude of reactions are mostly if not completely inactivated by the pasteurization temperature. A similar analysis approach was attempted for chocolate analysis. Here flavour differences were not as clearly recognised as for the cheese samples. In chocolate there are a greater number of parameters that are involved in its manufacture. Nevertheless, some recognisable differences in chocolate could be attributed to cocoa bean type and flavour additions by the manufacturer.

Real time detection of Volatile Organic Compounds (VOCs) using Selected Ion Flow Tube – Mass Spectrometry (SIFT-MS) provides a unique opportunity for research into breath testing for clinical diagnosis. However, before engaging in research into breath analytes as markers of disease, appropriate breath collection methods are required. Collection of breath for SIFT-MS instruments fall into two categories, direct breath collection into the instrument and the remote breath collection onto a storage medium. This thesis describes the development and validation of both methods of breath collection equipment for SIFT-MS analysis. Development of the direct breath collection device involved standardising and optimising the way in which breath is sampled by SIFT-MS. Design considerations include ergonomics, patient safety, breathing resistance, materials, and appropriate operating conditions of the device. Results from materials testing showed that all materials emit VOCs and the best approach is to minimise VOC emission by careful material selection. To minimise flow resistance experienced by the patient, the capillary from which the SIFT-MS instrument samples, is placed as close as possible to the users mouth. The optimal operating temperature of the device was found to be 100°C - 120°C, which ensures that water vapour will not condense inside the capillary causing blockage. In order to ensure patient safety the device is adequately insulated using stagnant air which also minimises VOC emission from insulation materials. Because a SIFT-MS instrument is large and cannot be easily shifted around a hospital, a system of remote sample collection is required. It is also important to separately collect and analyse breath from the respiratory alveolar region. For this reason the remote breath collection device designed also fractionates collected breath samples into the breath from the upper airways and alveolar breath. The storage medium chosen for the collected breath samples is a gas sampling bag made from Tedlar™. Collection of breath into Tedlar™ bags allows breath to be stored as a whole air sample, the ideal form for analysis with the SIFT-MS technique. Alveolar breath is fractionated from deadspace gasses by measuring a subject's exhalation and collecting the portion of interest. The breath exhalation is measured by an averaging Pitot tube and pressure transducer. Signal processing and automation of the remote breath collection device is controlled by a Cypress Microsystems PSoC microcontroller. To validate the device isoprene and acetone concentrations in fractionated breath samples were compared with a whole breath sample. Results showed that the alveolar breath fraction had a higher concentration of acetone than the upper airway fraction, indicating that the breath was successfully fractioned. However, isoprene concentrations were lower in both fractions due to hyperventilation of the subject causing a dilution effect of alveolar VOCs. Therefore, a higher sample collection volume is required per exhalation, and regulating subjects' breathing rate will avoid the dilution effect observed in collected breath samples. Overall, this thesis had designed, developed and validated two forms of breath collection systems for use with SIFT-MS technology.

In this paper, an object recognition system has been developed that uses local image features. In the system, multiple classes of objects can be recognized in an image. This system is basically divided into two parts: object detection and object identification. Object detection is based on SIFT features, which are invariant to image illumination, scaling and rotation. SIFT features extracted from a test image are used to perform a reliable matching between a database of SIFT features from known object images. Method of DBSCAN clustering is used for multiple object detection. RANSAC method is used for decreasing the amount of false detection. Object identification is based on 'Bag-of-Words' model. The 'BoW' model is a method based on vector quantization of SIFT descriptors of image patches. In this model, K-means clustering and Support Vector Machine (SVM) classification method are applied.

Made available in DSpace on 2014-06-12T15:55:49Z (GMT). No. of bitstreams: 2 arquivo2328_1.pdf: 4997289 bytes, checksum: 4012a5235a1aad082afe66cac56457eb (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2010 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / A reconstrução 3D é uma área de pesquisa que consiste em recuperar modelos que representem com precisão e em 3D características de interesse de uma cena, através da extração de informações 3D a partir de imagens 2D. Estas informações podem ser relativas à estrutura de uma determinada cena, posicionamento e trajetória de câmeras, textura, dentre outras. Uma vez de posse de tais informações, podemos utilizá-las para os mais diversos fins, por exemplo, modelagem automática de objetos, sistemas de navegação autônoma de robôs, modelos computacionais de estruturas ou órgãos do corpo humano, posicionamento de elementos virtuais em cenas reais, dentre outros. Uma das formas mais difundidas de se realizar reconstrução 3D é utilizando sequências contíguas de imagens ou vídeos capturados por câmeras convencionais (monoculares). Neste tipo de reconstrução um dos desafios mais importantes é o rastreamento. Rastreamento é a capacidade de conseguir corresponder um conjunto de pontos em uma sequência de imagens, ou seja, dado um ponto A com coordenadas x e y, deve-se ser capaz de identificar o ponto A com coordenadas x e y na imagem seguinte da sequência, e que corresponde exatamente à mesma localidade da estrutura sendo rastreada. Neste contexto, o objetivo desta dissertação de mestrado foi avaliar os algoritmos de rastreamento mais utilizados para este propósito, ressaltando as características individuais de cada um deles e identificando as vantagens e limitações que possuem. Os resultados desta análise podem ser uma ferramenta de auxílio na escolha do algoritmo de rastreamento a ser utilizado quando do desenvolvimento de uma solução de reconstrução 3D, tendo como base o domínio do problema que se deseja atacar. Os três algoritmos analisados foram o SIFT, o KLT e outro Baseado em Similaridade. Foi desenvolvida uma ferramenta de reconstrução 3D baseada em SfM. Esta ferramenta foi utilizada para a coleta de resultados com o rastreamento sendo realizado com SIFT, KLT e Similaridade. Uma etapa importante deste processo foi a definição de um conjunto de métricas para a análise comparativa dos algoritmos. As características individuais de rastreamento de cada um deles trouxeram bons resultados em alguns dos cinco cenários utilizados. Porém, no geral, o rastreador que apresentou os melhores resultados foi o KLT. Uma análise detalhada sobre os resultados desses algoritmos quando empregados para reconstrução 3D é apresentada

This thesis suggests a modification to the popular Scale Invariant Feature Transform (SIFT) algorithm (Lowe, 2004) often used in photogrammetry and computer vision to find features in images for measurement. The SIFT algorithm works by first detecting points in images at different scales and sizes. It then refines the position of the found points. The algorithm creates a descriptor of the point based on the region around the point. Finally the points can be matched against other points in different images using the descriptor. The suggested modification is built upon a paper by Förstner and Gülch (1987) where a method for performing a subwindow localization is presented. In this thesis the keypoints detected by the SIFT algorithm are modified on the subwindow level in order to improve the robustness with respect to the selected window position. Several different methods of tweaking the suggested modification and the SIFT algorithm were tested. The methods were evaluated on two different test cases. The first used a camera calibration software to compare accuracy of keypoints by looking at the residuals of the calibration. The other test involved creating a point cloud of images of a planar surface, evaluating the results by looking at the standard deviation in keypointoffset from the plane.The results show that neither test gave evidence that the proposed modification was an improvement. It was found that the algorithm had problems with oblique projections of circles, i.e. ellipses. Therefore there is potentialto use homography in special cases to circumvent this problem and get better precision. Furthermore tests involving more lines and intersections in the test images should be performed before this suggested modificationcan be completely discarded.

Context: In this thesis subimage matching in historical handwritten documents using SIFT (Scale-Invariant Feature Transform) keypoints was tested. SIFT features are invariant to scale and rotation and have gained a lot of interest in the research community. The historical documents used in this thesis orignates from 16th century and forward. The following steps have been executed; binarization, word segmentation, feature identification and clustering. The binarization step converts the images into binary images. The word segmentation separates the different words into individual subimages. In the feature identification SIFT keypoints was found and descriptors was computed. The last step was to cluster the images based on the distances between the set of image features identified. Objectives: The main objectives are to find a good configuration for the binarization step, implement a good word segmentation, identify image features and lastly to cluster the images based on their similarity. The context from subimages are matched to each other rather than trying to predict what the context of a subimage is, simply because the data that has been used is unlabeled. Methods: Implementation were the main methodology used combined with experimentation. Measurements were taken throughout the development and accuracy of word segmentation and the clustering is measured. Results: The word segmentation got an average accuracy of 89\% correct segmentation which is comparable to other word segmentating results. The clustering however matched 0% correctly.Conclusions: The conclusions that have been drawn from this study is that SIFT keypoints are not very well suited for this type of problem which includes a lot of handwritten text. The descriptors were not discriminative enough and different keypoints were found in different images with the same handwritten text, which lead to the bad clustering results.

No description available.

No description available.