Ένα σημαντικό πρόβλημα είναι η αντιστοίχιση εικόνων με σκοπό τη δημιουργία πανοράματος. Στην παρούσα εργασία έχουν χρησιμοποιηθεί αλγόριθμοι που βασίζονται στη χρήση σημείων κλειδιών. Αρχικά στην εργασία βρίσκονται σημεία κλειδιά για κάθε εικόνα που μένουν ανεπηρέαστα από τις αναμενόμενες παραμορφώσεις με την βοήθεια του αλγορίθμου SIFT (Scale Invariant Feature Transform). Έχοντας τελειώσει αυτή τη διαδικασία για όλες τις εικόνες, προσπαθούμε να βρούμε το πρώτο ζευγάρι εικόνων που θα ενωθεί. Για να δούμε αν δύο εικόνες μπορούν να ενωθούν, ακολουθεί ταίριασμα των σημείων κλειδιών τους. Όταν ένα αρχικό σετ αντίστοιχων χαρακτηριστικών έχει υπολογιστεί, πρέπει να βρεθεί ένα σετ που θα παράγει υψηλής ακρίβειας αντιστοίχιση. Αυτό το πετυχαίνουμε με τον αλγόριθμο RANSAC, μέσω του οποίου βρίσκουμε το γεωμετρικό μετασχηματισμό ανάμεσα στις δύο εικόνες, ομογραφία στην περίπτωσή μας. Αν ο αριθμός των κοινών σημείων κλειδιών είναι επαρκής, δηλαδή ταιριάζουν οι εικόνες, ακολουθεί η ένωσή τους. Αν απλώς ενώσουμε τις εικόνες, τότε θα έχουμε σίγουρα κάποια προβλήματα, όπως το ότι οι ενώσεις των δύο εικόνων θα είναι πολύ εμφανείς. Γι’ αυτό, για την εξάλειψη αυτού του προβλήματος, χρησιμοποιούμε τη μέθοδο των Λαπλασιανών πυραμίδων. Επαναλαμβάνεται η παραπάνω διαδικασία μέχρι να δημιουργηθεί το τελικό πανόραμα παίρνοντας κάθε φορά σαν αρχική την τελευταία εικόνα που φτιάξαμε στην προηγούμενη φάση. / Stitching multiple images together to create high resolution panoramas is one of the most popular consumer applications of image registration and blending. At this work, feature-based registration algorithms have been used. The first step is to extract distinctive invariant features from every image which are invariant to image scale and rotation, using SIFT (Scale Invariant Feature Transform) algorithm. After that, we try to find the first pair of images in order to stitch them. To check if two images can be stitched, we match their keypoints (the results from SIFT). Once an initial set of feature correspondences has been computed, we need to find the set that is will produce a high-accuracy alignment. The solution at this problem is RANdom Sample Consensus (RANSAC). Using this algorithm (RANSAC) we find the motion model between the two images (homography). If there is enough number of correspond points, we stitch these images. After that, seams are visible. As solution to this problem is used the method of Laplacian Pyramids. We repeat the above procedure using as initial image the ex panorama which has been created.

This work focuses on analysing and improving feature detection and matching. After creating an initial framework of study, four main areas of work are researched. These areas make up the main chapters within this thesis and focus on using the Scale Invariant Feature Transform (SIFT).The preliminary analysis of the SIFT investigates how this algorithm functions. Included is an analysis of the SIFT feature descriptor space and an investigation into the noise properties of the SIFT. It introduces a novel use of the a contrario methodology and shows the success of this method as a way of discriminating between images which are likely to contain corresponding regions from images which do not. Parameter analysis of the SIFT uses both parameter sweeps and genetic algorithms as an intelligent means of setting the SIFT parameters for different image types utilising a GPGPU implementation of SIFT. The results have demonstrated which parameters are more important when optimising the algorithm and the areas within the parameter space to focus on when tuning the values. A multi-exposure, High Dynamic Range (HDR), fusion features process has been developed where the SIFT image features are matched within high contrast scenes. Bracketed exposure images are analysed and features are extracted and combined from different images to create a set of features which describe a larger dynamic range. They are shown to reduce the effects of noise and artefacts that are introduced when extracting features from HDR images directly and have a superior image matching performance. The final area is the development of a novel, 3D-based, SIFT weighting technique which utilises the 3D data from a pair of stereo images to cluster and class matched SIFT features. Weightings are applied to the matches based on the 3D properties of the features and how they cluster in order to attempt to discriminate between correct and incorrect matches using the a contrario methodology. The results show that the technique provides a method for discriminating between correct and incorrect matches and that the a contrario methodology has potential for future investigation as a method for correct feature match prediction.

Nous étudions ici l’intérêt des descripteurs locaux pour les images satellites optiques et radar. Ces descripteurs, par leurs invariances et leur représentation compacte, offrent un intérêt pour la comparaison d’images acquises dans des conditions différentes. Facilement applicables aux images optiques, ils offrent des performances limitées sur les images radar, en raison de leur fort bruit multiplicatif. Nous proposons ici un descripteur original pour la comparaison d’images radar. Cet algorithme, appelé SAR-SIFT, repose sur la même structure que l’algorithme SIFT (détection de points-clés et extraction de descripteurs) et offre des performances supérieures pour les images radar. Pour adapter ces étapes au bruit multiplicatif, nous avons développé un opérateur différentiel, le Gradient par Ratio, permettant de calculer une norme et une orientation du gradient robustes à ce type de bruit. Cet opérateur nous a permis de modifier les étapes de l’algorithme SIFT. Nous présentons aussi deux applications pour la télédétection basées sur les descripteurs. En premier, nous estimons une transformation globale entre deux images radar à l’aide de SAR-SIFT. L’estimation est réalisée à l’aide d’un algorithme RANSAC et en utilisant comme points homologues les points-clés mis en correspondance. Enfin nous avons mené une étude prospective sur l’utilisation des descripteurs pour la détection de changements en télédétection. La méthode proposée compare les densités de points-clés mis en correspondance aux densités de points-clés détectés pour mettre en évidence les zones de changement. / We study here the interest of local features for optical and SAR images. These features, because of their invariances and their dense representation, offer a real interest for the comparison of satellite images acquired under different conditions. While it is easy to apply them to optical images, they offer limited performances on SAR images, because of their multiplicative noise. We propose here an original feature for the comparison of SAR images. This algorithm, called SAR-SIFT, relies on the same structure as the SIFT algorithm (detection of keypoints and extraction of features) and offers better performances for SAR images. To adapt these steps to multiplicative noise, we have developed a differential operator, the Gradient by Ratio, allowing to compute a magnitude and an orientation of the gradient robust to this type of noise. This operator allows us to modify the steps of the SIFT algorithm. We present also two applications for remote sensing based on local features. First, we estimate a global transformation between two SAR images with help of SAR-SIFT. The estimation is realized with help of a RANSAC algorithm and by using the matched keypoints as tie points. Finally, we have led a prospective study on the use of local features for change detection in remote sensing. The proposed method consists in comparing the densities of matched keypoints to the densities of detected keypoints, in order to point out changed areas.

La profondeur possible d’imagerie en laser-scanning microscopie est limitée non seulement par la distance de travail des lentilles de objectifs mais également par la dégradation de l’image causée par une atténuation et une diffraction de la lumière passant à travers l’échantillon. Afin d’étendre cette limite, il est possible, soit de retourner le spécimen pour enregistrer les images depuis chaque côté, or couper progressivement la partie supérieure de l’échantillon au fur et à mesure de l‘acquisition. Les différentes images prises de l’une de ces manières doivent ensuite être combinées pour générer un volume unique. Cependant, des mouvements de l’échantillon durant les procédures d’acquisition engendrent un décalage non seulement sur en translation selon les axes x, y et z mais également en rotation autour de ces même axes, rendant la fusion entres ces multiples images difficile. Nous avons développé une nouvelle approche appelée 2D-SIFT-in-3D-Space utilisant les SIFT (scale Invariant Feature Transform) pour atteindre un recalage robuste en trois dimensions de deux images. Notre méthode recale les images en corrigeant séparément les translations et rotations sur les trois axes grâce à l’extraction et l’association de caractéristiques stables de leurs coupes transversales bidimensionnelles. Pour évaluer la qualité du recalage, nous avons également développé un simulateur d’images de laser-scanning microscopie qui génère une paire d’images 3D virtuelle dans laquelle le niveau de bruit et les angles de rotations entre les angles de rotation sont contrôlés avec des paramètres connus. Pour une concaténation précise et naturelle de deux images, nous avons également développé un module permettant une compensation progressive de la luminosité et du contraste en fonction de la distance à la surface de l’échantillon. Ces outils ont été utilisés avec succès pour l’obtention d’images tridimensionnelles de haute résolution du cerveau de la mouche Drosophila melanogaster, particulièrement des neurones dopaminergiques, octopaminergiques et de leurs synapses. Ces neurones monoamines sont particulièrement important pour le fonctionnement du cerveau et une étude de leur réseau et connectivité est nécessaire pour comprendre leurs interactions. Si une évolution de leur connectivité au cours du temps n’a pas pu être démontrée via l’analyse de la répartition des sites synaptiques, l’étude suggère cependant que l’inactivation de l’un de ces types de neurones entraine des changements drastiques dans le réseau neuronal. / Although laser scanning microscopy is a powerful tool for obtaining thin optical sections, the possible depth of imaging is limited by the working distance of the microscope objective but also by the image degradation caused by the attenuation of both excitation laser beam and the light emitted from the fluorescence-labeled objects. Several workaround techniques have been employed to overcome this problem, such as recording the images from both sides of the sample, or by progressively cutting off the sample surface. The different views must then be combined in a unique volume. However, a straightforward concatenation is often not possible, because the small rotations that occur during the acquisition procedure, not only in translation along x, y and z axes but also in rotation around those axis, making the fusion uneasy. To address this problem we implemented a new algorithm called 2D-SIFT-in-3D-Space using SIFT (scale Invariant Feature Transform) to achieve a robust registration of big image stacks. Our method register the images fixing separately rotations and translations around the three axes using the extraction and matching of stable features in 2D cross-sections. In order to evaluate the registration quality, we created a simulator that generates artificial images that mimic laser scanning image stacks to make a mock pair of image stacks one of which is made from the same stack with the other but is rotated arbitrarily with known angles and filtered with a known noise. For a precise and natural-looking concatenation of the two images, we also developed a module progressively correcting the sample brightness and contrast depending on the sample surface. Those tools we successfully used to generate tridimensional high resolution images of the fly Drosophila melanogaster brain, in particular, its octopaminergic and dopaminergic neurons and their synapses. Those monoamine neurons appear to be determinant in the correct operating of the central nervous system and a precise and systematic analysis of their evolution and interaction is necessary to understand its mechanisms. If an evolution over time could not be highlighted through the pre-synaptic sites analysis, our study suggests however that the inactivation of one of these neuron types triggers drastic changes in the neural network.

In the area of automation, object localization stands as a fundamental functionalitywith widespread applicability. This master’s thesis focuses on a specificapplication involving robot object picking. Given recent advancements in depthcamera technology, there is a high interest in exploring the synergistic integrationof both 2D and 3D data to address challenges such as missing data, occlusion,varying viewing angles, and diverse lighting conditions. This master’s thesis presents the development of two distinct algorithms for arbitraryshaped template matching using 2D image features. Both algorithms leveragefeatures detected by the GoodFeaturesToTrack algorithm and described withScale-invariant feature transform (SIFT) descriptors. While an initial sliding windowmatcher was developed, it was ultimately discarded due to extensive timerequirements. Instead, a correspondence matcher was created, offering two variations:one exclusively employing 2D image data for matching and another utilizing3D coordinates to enhance matching accuracy. The correspondence matchingalgorithms showed similar strengths and weaknesses. They demonstrated proficiencyin handling scenarios characterized by occlusion, minor tilt, and varyingscaling. Both variations struggled with objects 90-degrees rotated and could inmany cases not find them. The findings suggest that the developed feature-based correspondence matchingalgorithm holds promise for object localization in industrial picking applications,although with limitations concerning objects with substantial rotationdifferences. Addressing the challenge of large rotations is recommended for enhancingthe algorithm’s robustness, followed by comprehensive testing to ascertainits efficacy in diverse scenarios.iii

This thesis presents the novel use of Coherent Point Drift in improving the robustness of a number of computer vision applications. CPD approach includes two methods for registering two images - rigid and non-rigid point set approaches which are based on the transformation model used. The key characteristic of a rigid transformation is that the distance between points is preserved, which means it can be used in the presence of translation, rotation, and scaling. Non-rigid transformations - or affine transforms - provide the opportunity of registering under non-uniform scaling and skew. The idea is to move one point set coherently to align with the second point set. The CPD method finds both the non-rigid transformation and the correspondence distance between two point sets at the same time without having to use a-priori declaration of the transformation model used. The first part of this thesis is focused on speaker identification in video conferencing. A real-time, audio-coupled video based approach is presented, which focuses more on the video analysis side, rather than the audio analysis that is known to be prone to errors. CPD is effectively utilised for lip movement detection and a temporal face detection approach is used to minimise false positives if face detection algorithm fails to perform. The second part of the thesis is focused on multi-exposure and multi-focus image fusion with compensation for camera shake. Scale Invariant Feature Transforms (SIFT) are first used to detect keypoints in images being fused. Subsequently this point set is reduced to remove outliers, using RANSAC (RANdom Sample Consensus) and finally the point sets are registered using CPD with non-rigid transformations. The registered images are then fused with a Contourlet based image fusion algorithm that makes use of a novel alpha blending and filtering technique to minimise artefacts. The thesis evaluates the performance of the algorithm in comparison to a number of state-of-the-art approaches, including the key commercial products available in the market at present, showing significantly improved subjective quality in the fused images. The final part of the thesis presents a novel approach to Vehicle Make & Model Recognition in CCTV video footage. CPD is used to effectively remove skew of vehicles detected as CCTV cameras are not specifically configured for the VMMR task and may capture vehicles at different approaching angles. A LESH (Local Energy Shape Histogram) feature based approach is used for vehicle make and model recognition with the novelty that temporal processing is used to improve reliability. A number of further algorithms are used to maximise the reliability of the final outcome. Experimental results are provided to prove that the proposed system demonstrates an accuracy in excess of 95% when tested on real CCTV footage with no prior camera calibration.

Availability of digital cameras in various forms and user-friendly image editing softwares has enabled people to create and manipulate digital images easily. While image editing can be used for enhancing the quality of the images, it can also be used to tamper the images for malicious purposes. In this context, it is important to question the originality of digital images. Digital image forensics deals with the development of algorithms and systems to detect tampering in digital images. This thesis presents some simple algorithms which can be used to detect tampering in digital images. Out of the various kinds of image forgeries possible, the discussion is restricted to photo compositing (Photo montaging) and copy-paste forgeries. While creating photomontage, it is very likely that one of the images needs to be resampled and hence there will be an inconsistency in some of its underlying characteristics. So, detection of resampling in an image will give a clue to decide whether the image is tampered or not. Two pixel domain techniques to detect resampling have been presented. The rest of them exploits the property of periodic zeros that occur in the second divergences due to interpolation during resembling. It requires a special condition on the resembling factor to be met. The second technique is based on the periodic zero-crossings that occur in the second divergences, which does not require any special condition on the resembling factor. It has been noted that this is an important property of revamping and hence the decay of this technique against mild counter attacks such as JPEG compression and additive noise has been studied. This property has been repeatedly used throughout this thesis. It is a well known fact that interpolation is essentially low-pass filtering. In case of photomontage image which consists of resample and non resample portions, there will be an in consistency in the high-frequency content of the image. This can be demonstrated by a simple high-pass filtering of the image. This fact has also been exploited to detect photomontaging. One approach involves performing block wise DCT and reconstructing the image using only a few high-frequency coercions. Another elegant approach is to decompose the image using wavelets and reconstruct the image using only the diagonal detail coefficients. In both the cases mere visual inspection will reveal the forgery. The second part of the thesis is related to tamper detection in colour filter array (CFA) interpolated images. Digital cameras employ Bayer filters to efficiently capture the RGB components of an image. The output of Bayer filter are sub-sampled versions of R, G and B components and they are completed by using demosaicing algorithms. It has been shown that demos icing of the color components is equivalent to resembling the image by a factor of two. Hence, CFA interpolated images contain periodic zero-crossings in its second differences. Experimental demonstration of the presence of periodic zero-crossings in images captured using four digital cameras of deferent brands has been done. When such an image is tampered, these periodic zero-crossings are destroyed and hence the tampering can be detected. The utility of zero-crossings in detecting various kinds of forgeries on CFA interpolated images has been discussed. The next part of the thesis is a technique to detect copy-paste forgery in images. Generally, while an object or a portion if an image has to be erased from an image, the easiest way to do it is to copy a portion of background from the same image and paste it over the object. In such a case, there are two pixel wise identical regions in the same image, which when detected can serve as a clue of tampering. The use of Scale-Invariant-Feature-Transform (SIFT) in detecting this kind of forgery has been studied. Also certain modifications that can to be done to the image in order to get the SIFT working effectively has been proposed. Throughout the thesis, the importance of human intervention in making the final decision about the authenticity of an image has been highlighted and it has been concluded that the techniques presented in the thesis can effectively help the decision making process.

This thesis describes methods of reconstruction of 3D scenes from photographs and videos using the Structure from motion approach. A new software capable of automatic reconstruction of point clouds and polygonal models from common images and videos was implemented based on these methods. The software uses variety of existing and custom solutions and clearly links them into one easily executable application. The reconstruction consists of feature point detection, pairwise matching, Bundle adjustment, stereoscopic algorithms and polygon model creation from point cloud using PCL library. Program is based on Bundler and PMVS. Poisson surface reconstruction algorithm, as well as simple triangulation and own reconstruction method based on plane segmentation were used for polygonal model creation.