In this thesis, rotation, scale and translation (RST) invariant automatic target recognition (ATR) for satellite imagery is presented. Template matching is used to realize the target recognition. However, unlike most of the studies of template matching in the literature, RST invariance is required in our problem, since most of the time we will have only a small number of templates of each target, while the targets to be recognized in the scenes will have various orientations, scaling and translations. RST invariance is studied in detail and implemented with some of the competing methods in the literature, such as Fourier-Mellin transform and bipectrum combined with log-polar mapping. Phase correlation and normalized cross-correlation are used as similarity metrics. Encountered drawbacks were overcome with additional operations and modifications of the algorithms. ATR using reconstruction of the target image with respect to the template, based on bispectrum, log-polar mapping and phase correlation outperformed the other methods and successful recognition was realized for various target types, especially for targets on relatively simpler backgrounds, i.e. containing little or no other objects.

Road extraction from satellite or aerial images is a popular issue in remote sensing. Extracted road maps or networks can be used in various applications. Normally, maps for roads are available in geographic information systems (GIS), however these informations are not being produced automatically. Generally they are formed with the aid of human. Road extraction algorithms are trying to detect the roads from satellite or aerial images with the minimum in-teraction of human. Aim of this thesis is to analyze a previously defined algorithm about road extraction and to present alternatives and possible improvements to this algorithm. The base-line algorithm and proposed alternative algorithm and steps are based on mean-shift segmen-tation procedure. Proposed alternative methods are generally based on structural features of the image. Firstly, fundamental definitions of applied algorithms and methods are explained, mathematical definitions and visual examples are given for better understanding. Then, the chosen baseline algorithm and its alternatives are explained in detail. After the presentation of alternative methods, experimental results and inferences which are obtained during the implementation and analysis of mentioned algorithms and methods are presented.

Dynamic speech properties such as time warping, silence removal and background noise interference are the most challenging issues in continuous speech signal matching. Among all of them, the time warped speech signal matching is of great interest and has been a tough challenge for the researchers. An adaptive framing based continuous speech tracking and similarity measurement approach is introduced in this work following a comprehensive research conducted in the diverse areas of speech processing. A dynamic state model is introduced based on system of linear motion equations which models the input (test) speech signal frame as a unidirectional moving object along the template speech signal. The most similar corresponding frame position in the template speech is estimated which is fused with a feature based similarity observation and the noise variances using a Kalman filter. The Kalman filter provides the final estimated frame position in the template speech at current time which is further used for prediction of a new frame size for the next step. In addition, a keyword spotting approach is proposed by introducing wavelet decomposition based dynamic noise filter and combination of beliefs. The Dempster’s theory of belief combination is deployed for the first time in relation to keyword spotting task. Performances for both; speech tracking and keyword spotting approaches are evaluated using the statistical metrics and gold standards for the binary classification. Experimental results proved the superiority of the proposed approaches over the existing methods.

Bien que devenant un environnement de plus en plus riche technologiquement, la salle opératoire reste un endroit où la sécurité des patients n'est pas assurée à 100% comme le montre le nombre toujours conséquent d'erreurs chirurgicales. La nécessité de développer des systèmes intelligents au bloc opératoire apparait comme croissante. Un des éléments clés pour ce type de système est la reconnaissance du processus chirurgical, passant par une identification précise des outils chirurgicaux utilisés. L'objectif de cette thèse a donc porté sur la détection en temps-réel des outils chirurgicaux dans des vidéos 2D provenant de microscopes. Devant l'absence de jeux de données de référence, qui plus est dans un contexte neurochirurgical, la première contribution de la thèse a donc été la création d'un nouvel ensemble d'images de chirurgies du cerveau et du rachis cervical, mis à disposition en ligne. Comme seconde contribution, deux approches différentes ont été proposées permettant de détecter des outils chirurgicaux via des techniques d'analyse d'image. Tout d'abord, le SquaresChnFtrs adapté, basé sur une des méthodes de détection de piétons les plus performantes de la littérature. Notre deuxième méthode, le ShapeDetector, est une approche à deux niveaux n'utilisant aucune contrainte ou hypothèse a priori sur le nombre, la position, ou la forme des outils dans l'image. Par rapport aux travaux précédents du domaine, nous avons choisi de représenter les détections potentielles par des polygones plutôt que par des rectangles, obtenant ainsi des détections plus précises. Pour intégration dans des systèmes médicaux, une optimisation de la vitesse de calcul a été effectuée via un usage optimal du CPU, du GPU, et de méthodes ad-hoc. Pour des vidéos de résolution 612x480 pixels, notre ShapeDetector est capable d'effectuer les détections à une vitesse maximale de 8 Hz. Pour la validation de nos méthodes, nous avons pris en compte trois paramètres: la position globale, la position de l'extrémité, et l'orientation des détections. Les méthodes ont été classées et comparées avec des méthodes de référence compétitives. Pour la détection des tubes d'aspiration, nous avons obtenu un taux de manqué de 15% pour un taux de faux positifs par image de 0.1, comparé à un taux de manqué de 55% pour le SquaresChnFtrs adapté. L'orientation future du travail devra porter sur l'intégration des informations 3D, l'amélioration de la couche de labélisation sémantique, et la classification des outils chirurgicaux. Pour finir, un enrichissement du jeu de données et des annotations de plus haute précision seront nécessaires. / Despite modern-life technological advances and tremendous progress made in surgical techniques including MIS, today's OR is facing many challenges remaining yet to be addressed. The development of CAS systems integrating the SPM methodology was born as a response from the medical community, with the long-term objective to create surgical cockpit systems. Being able to identify surgical tools in use is a key component for systems relying on the SPM methodology. Towards that end, this thesis work has focused on real-time surgical tool detection from microscope 2D images. From the review of the literature, no validation data-sets have been elected as benchmarks by the community. In addition, the neurosurgical context has been addressed only once. As such, the first contribution of this thesis work consisted in the creation of a new surgical tool data-set, made freely available online. Two methods have been proposed to tackle the surgical tool detection challenge. First, the adapted SquaresChnFtrs, evolution of one of the best computer vision state-of-the-art approach for pedestrian detection. Our second contribution, the ShapeDetector, is fully data-driven and performs detection without the use of prior knowledge regarding the number, shape, and position of tools in the image. Compared to previous works, we chose to represent candidate detections with bounding polygons instead of bounding boxes, hence providing more fitting results. For integration into medical systems, we performed different code optimization through CPU and GPU use. Speed gain and accuracy loss from the use of ad-hoc optimization strategies have been thoroughly quantified to find an optimal trade-off between speed and accuracy. Our ShapeDetector is running in-between 5 and 8Hz for 612x480 pixel video sequences.We validated our approaches using a detailed methodology covering the overall tool location, tip position, and orientation. Approaches have been compared and ranked conjointly with a set of competitive baselines. For suction tube detections, we achieved a 15% miss-rate at 0.1 FPPI, compared to a 55% miss-rate for the adapted SquaresChnFtrs. Future works should be directed toward the integration of 3D feature extraction to improve detection performance but also toward the refinement of the semantic labelling step. Coupling the tool detection task to the tool classification in one single framework should be further investigated. Finally, increasing the data-set in diversity, number of tool classes, and detail of annotations is of interest.

This thesis focuses on the development of an open source software capable of automatic face detection in an image captured by a thermal camera, followed by a temperature measuring. This software is supposed to aid in the COVID-19 pandemics. The developed software is independent of used thermal camera. In this thesis, I am using TIM400 thermal camera. The implementation of the face detection was achieved by an OpenCV module. The methods tested were Template Matching, Eigen Faces, and Cascade Classifier. The last-mentioned had the best results, hence was used in the final version of the software. Cascade Classifier is looking for the eyes and their surrounding area in the image, allowing the software to subsequently measure the temperature on the surface of one's forehead. One can therefore be wearing a face mask or a respirator safely. The temperature measuring works in real time and the software is able to capture several people at once. It then keeps a record of the temperature of each measured individual as well as the time of the measurement. The software as a whole is a part of an installation file compatible with the Windows operating system. The functionality of this software was tested – the video recordings are included in this thesis.

Quality control is an important part of any production line. It can be done manually but is most efficient if automated. Inspecting qualitycan include many different processes but this thesisisfocusedon the visual inspection for cracks and scratches. The best way of doingthis at the time of writing is with the help of Artificial Intelligence (AI), more specifically Deep Learning (DL).However, these need a training datasetbeforehand to train on and for some smaller companies, this mightnotbean option. This study triesto find an alternative visual inspection method,that does notrelyon atrained deep learning modelfor when trainingdata is severely limited. Our method is to use edge detection algorithmsin combination with a template to find any edge that doesn’t belong. These include scratches, cracks, or misaligned stickers. These anomalies arethen highlighted in the original picture to show where the defect is. Since deep learningis stateof the art ofvisual inspection, it is expected to outperform template matching when sufficiently trained.To find where this occurs,the accuracy of template matching iscompared to the accuracy of adeep learning modelat different training levels. The deep learning modelisto be trained onimage augmenteddatasets of size: 6, 12, 24, 48, 84, 126, 180, 210, 315, and 423. Both template matching and the deep learning modelwas tested on the samebalanceddataset of size 216. Half of the dataset was images of scratched units,and the other half was of unscratched units. This gave a baseline of 50% where anything under would be worse thanjust guessing. Template matching achieved an accuracy of 88%, and the deep learning modelaccuracyrose from 51% to 100%as the training setincreased. This makes template matching have better accuracy then AI trained on dataset of 84imagesor smaller. But a deep learning modeltrained on 126 images doesstart to outperform template matching. Template matching did perform well where no data was available and training adeep learning modelis no option. But unlike a deep learning model, template matching would not need retraining to find other kinds of surface defects. Template matching could also be used to find for example, misplaced stickers. Due to the use of a template, any edge that doesnot match isdetected.  The ways to train deep learning modelis highly customizable to the users need. Due to resourceand knowledge restrictions, a deep dive into this subject was not conducted.For template matching, only Canny edge detection was used whenmeasuringaccuracy. Other edge detection methodssuch as, Sobel, and Prewitt was ruledoutearlier in this study.

This thesis deal with various methods of background detection and with it related motion detection in a scene. It's progressing from simplest methods to more comlex. For every one are reviewed the possibilities of using and her drawbacks. In introduction are described various types of scenes according to background and foreground type e.g . according to movement objects speed or presence of movement in background. Is proposed several common or specific improvements for obtaining better background even by using simple method. Next part of work solve real situation of shaking camera. There are tested two basic methods for optical stabilization. The first is registration of images by template matching. Alternative method used interest points (corners). Both methods are closely examinate and is sought best way to match following pictures. Except shaking of camera this work deal with rotating camera and in theory solve detection background from cameras placed on ridden car. Part of work is creation database of different types scenes

Developing the greenest battery cell and establishing a European supply of batteries is the main goal for Northvolt. To achieve this, the recycling of batteries is a key enabler towards closing the loop and enabling the future of energy.When it comes to the recycling of electric vehicle battery packs, dismantling is one of of the main process steps.Given the size, weight and high voltage of the battery packs, automatic disassembly using robots is the preferred solution. The work presented in this thesis aims to develop and integrate a vision system able to identify and verify the battery pack dismantling process. To achieve this, two cameras were placed in the robot cell and the object detectors You Only Look Once (YOLO) and template matching were implemented, tested and compared. The results show that YOLO is the best object detector out of the ones implemented. The integration of the vision system with the robot controller was also tested and showed that with the results from the vision system, the robot controller can make informed decisions regarding the disassembly.

Internet of Things (IoT) are becoming increasingly popular replacements for their analogue counterparts. However, there is still demand to keep analogue equipment that is already installed, while also having automated monitoring of the equipment, such as analogue water meters. A proposed solution for this problem is to install a battery powered add-on component that can optically read meter values using Optical Character Recognition (OCR) and transmit the readings wirelessly. Two ways to do this could be to either offload the OCR process to a server, or to do the OCR processing locally on the add-on component. Since water meters are often located where reception is weak and the add-on component is battery powered, a suitable technology for data transmission could be Long Range (LoRa) because of its low-power and long-range capabilities. Since LoRa has low transfer rate there is a need to keep data transfers small in size, which could make offloading a less favorable alternative compared to local OCR processing. The purpose of this thesis is therefore to research the feasibility, in terms of energy efficiency, of doing local OCR processing on the add-on component. The feasibility condition of this study is defined as being able to continually read an analogue meter for a 10-year lifespan, while consuming under 2600 milliampere hours (mAh) of energy. The two OCR algorithms developed for this study are a specialized OCR algorithm that utilizes pattern matching principles, and a Sum of Absolute Differences (SAD) OCR algorithm. These two algorithms have been compared against each other, to determine which one is more suitable for the system. This comparison yielded that the SAD algorithm was more suitable, and was then studied further by using different image resolutions and settings to determine if it was possible to further reduce energy consumption. The results showed that it was possible to significantly reduce energy consumption by reducing the image resolution. The study also researched the possibility of reducing energy consumption further by not reading all digits on the tested water meter, depending on the measuring frequency and water flow. The study concluded that OCR processing is feasible on an energy critical driven system when reading analouge meters, depending on the measuring frequency.

The field of design automation aims to automate repetitive tasks in a workflow in order to free up time for more productive work. In this thesis, design automation with the help of AI techniques is investigated to streamline the pre-production work of expanded metal facades.  Two different problems concerning pre-production work are investigated in this thesis. The first one focuses on how to translate architectural drawings in pdf format to a bill of material. The second problem aims to develop a non-linear method for calculating the free area of the expanded metal facades. The method used for this project is an adaptation of the product development process with the inspiration of knowledge-based engineering.  For the first project, the AI method template matching was successfully used. With a script using this method, most of the panels are identified, except for panels where the drawings do not provide clear lines or where lines around the panels do not exist. The line quality in the architectural drawings was shown to impact the size estimation of the panels. In the second project, a non-linear machine learning model was developed. However, it was not managed within this project to get a good enough accuracy. The main reason for this is that it is suspected that the data is not accurate enough, nor are the 78 data points enough to train the model.