Zernike moments are sets of mathematical quantities that uniquely characterize an image. It is known that they are invariant under rotation and reflection and robust to noise. In this thesis several other algorithms have been used to calculate these moments. The intent of this thesis is: 1. to use the classical method and the algorithms to reconstruct an image using Zernike moments and study their accuracy and 2. to examine if the invariance and noise insensitivity property of the calculated Zernike moments are upheld by these procedures. It is found that the constructed images using these algorithms do not resemble the original image. This prevents us from carrying out further study of these algorithms. The classical method has been successfully used to reconstruct an image when the height and width are equal. The classical method is also shown to be invariant under rotation and reflection and robust to Poisson noise. xxxvii

Zernike moments are sets of mathematical quantities that uniquely characterize an image. It is known that they are invariant under rotation and reflection and robust to noise. In this thesis several other algorithms have been used to calculate these moments. The intent of this thesis is: 1. to use the classical method and the algorithms to reconstruct an image using Zernike moments and study their accuracy and 2. to examine if the invariance and noise insensitivity property of the calculated Zernike moments are upheld by these procedures. It is found that the constructed images using these algorithms do not resemble the original image. This prevents us from carrying out further study of these algorithms. The classical method has been successfully used to reconstruct an image when the height and width are equal. The classical method is also shown to be invariant under rotation and reflection and robust to Poisson noise. xxxvii

The 3-D fluorescence microscope has become the method of choice in biological sciences for living cells study. However, the data acquired with conventional3-D fluorescence microscopy are not quantitatively significant because of distortions induced by the optical acquisition process. Reliable measurements need the correction of theses distortions. Knowing the instrument impulse response, also known as the PSF, one can consider the backward process of convolution induced by the microscope, known as "deconvolution". However, when the system response is not invariant in the observation field, the classical algorithms can introduce large errors in the results. In this thesis we propose a new approach, which can be easily adapted to any classical deconvolution algorithm, direct or iterative, for bypassing the non-invariance PSF problem, without any modification to the later. Based on the hypothesis that the minimal error in a restored image using non-invariance assumption is located near the used PSF position, the EMMA (Evolutive Merging Masks Algorithm) blends multiple deconvolutions in the invariance assumption using a specific merging mask set. In order to obtain sufficient number of measured PSF at various depths for a better restoration using EMMA (or any other depth-variant deconvolution algorithm) we propose a 3D PSF interpolation algorithm based on the image moments theory using Zernike polynomials as decomposition base. The known PSF are decomposed into Zernike moments set and each moment's variation is fitted into a polynomial function, the resulting functions are then used to interpolate the needed PSF's Zernike moments set to reconstruct the interpolated PSF.

This thesis presents a novel lip-reading approach to classifying utterances from video data, without evaluating voice signals. This work addresses two important issues which are • the efficient representation of mouth movement for visual speech recognition • the temporal segmentation of utterances from video. The first part of the thesis describes a robust movement-based technique used to identify mouth movement patterns while uttering phonemes. This method temporally integrates the video data of each phoneme into a 2-D grayscale image named as a motion template (MT). This is a view-based approach that implicitly encodes the temporal component of an image sequence into a scalar-valued MT. The data size was reduced by extracting image descriptors such as Zernike moments (ZM) and discrete cosine transform (DCT) coefficients from MT. Support vector machine (SVM) and hidden Markov model (HMM) were used to classify the feature descriptors. A video speech corpus of 2800 utterances was collected for evaluating the efficacy of MT for lip-reading. The experimental results demonstrate the promising performance of MT in mouth movement representation. The advantages and limitations of MT for visual speech recognition were identified and validated through experiments. A comparison between ZM and DCT features indicates that th e accuracy of classification for both methods is very comparable when there is no relative motion between the camera and the mouth. Nevertheless, ZM is resilient to rotation of the camera and continues to give good results despite rotation but DCT is sensitive to rotation. DCT features are demonstrated to have better tolerance to image noise than ZM. The results also demonstrate a slight improvement of 5% using SVM as compared to HMM. The second part of this thesis describes a video-based, temporal segmentation framework to detect key frames corresponding to the start and stop of utterances from an image sequence, without using the acoustic signals. This segmentation technique integrates mouth movement and appearance information. The efficacy of this technique was tested through experimental evaluation and satisfactory performance was achieved. This segmentation method has been demonstrated to perform efficiently for utterances separated with short pauses. Potential applications for lip-reading technologies include human computer interface (HCI) for mobility-impaired users, defense applications that require voice-less communication, lip-reading mobile phones, in-vehicle systems, and improvement of speech-based computer control in noisy environments.

This thesis presents a technique for face recognition that is based on image registration. The image registration technique is based on finding a set of feature points in the two images and using these feature points for registration. This is done in four steps. In the first, images are filtered with the Mexican hat wavelet to obtain the feature point locations. In the second, the Zernike moments of neighbourhoods around the feature points are calculated and compared in the third step to establish correspondence between feature points in the two images and in the fourth the transformation parameters between images are obtained using an iterative weighted least squares technique. The face recognition technique consists of three parts, a training part, an image registration part and a post-processing part. During training a set of images are chosen as the training images and the Zernike moments for the feature points of the training images are obtained and stored. In the registration part, the transformation parameters to register the training images with the images under consideration are obtained. In the post-processing, these transformation parameters are used to determine whether a valid match is found or not. The performance of the proposed method is evaluated using various face databases and it is compared with the performance of existing techniques. Results indicate that the proposed technique gives excellent results for face recognition in conditions of varying pose, illumination, background and scale. These results are comparable to other well known face recognition techniques.

La reconstruction 3D par coupes sériées en microscopie optique est un moyen efficace pour étudier des spécimens biologiques fluorescents. Dans un tel système, la formation d'une image peut être représentée comme une convolution linéaire d'un objet avec une réponse impulsionnelle optique de l'instrument (PSF). Pour une étude quantitative, une estimation de l'objet doit être calculée en utilisant la déconvolution qui est le phénomène inverse de la convolution. Plusieurs algorithmes de déconvolution ont été développés en se basant sur des modèles statistiques ou par inversion directe, mais ces algorithmes se basent sur la supposition de l'invariance spatiale de la PSF pour simplifier et accélérer le processus. Dans certaines configurations optiques la PSF 3D change significativement en profondeur et ignorer ces changements implique des erreurs quantitatives dans l'estimation. Nous proposons un algorithme (EMMA) qui se base sur une hypothèse où l'erreur minimale sur l'estimation par un algorithme ne tenant pas compte de la non-invariance, se situe aux alentours de la position (profondeur) de la PSF utilisée. EMMA utilise des PSF à différentes positions et fusionne les différentes estimations en utilisant des masques d'interpolation linéaires adaptatifs aux positions des PSF utilisées. Pour obtenir des PSF à différentes profondeurs, un algorithme d'interpolation de PSF a également été développé. La méthode consiste à décomposer les PSF mesurées en utilisant les moments de Zernike pseudo-3D, puis les variations de chaque moment sont approximés par une fonction polynomiale. Ces fonctions polynomiales sont utilisées pour interpoler des PSF aux profondeurs voulues. / The 3-D fluorescence microscope has become the method of choice in biological sciences for living cells study. However, the data acquired with conventional3-D fluorescence microscopy are not quantitatively significant because of distortions induced by the optical acquisition process. Reliable measurements need the correction of theses distortions. Knowing the instrument impulse response, also known as the PSF, one can consider the backward process of convolution induced by the microscope, known as "deconvolution". However, when the system response is not invariant in the observation field, the classical algorithms can introduce large errors in the results. In this thesis we propose a new approach, which can be easily adapted to any classical deconvolution algorithm, direct or iterative, for bypassing the non-invariance PSF problem, without any modification to the later. Based on the hypothesis that the minimal error in a restored image using non-invariance assumption is located near the used PSF position, the EMMA (Evolutive Merging Masks Algorithm) blends multiple deconvolutions in the invariance assumption using a specific merging mask set. In order to obtain sufficient number of measured PSF at various depths for a better restoration using EMMA (or any other depth-variant deconvolution algorithm) we propose a 3D PSF interpolation algorithm based on the image moments theory using Zernike polynomials as decomposition base. The known PSF are decomposed into Zernike moments set and each moment's variation is fitted into a polynomial function, the resulting functions are then used to interpolate the needed PSF's Zernike moments set to reconstruct the interpolated PSF.