Nouryon, formerly Akzo Nobel, is a world-leading manufacturer of particles. Already in the 70s, one part of Nouryon was founded in Stockvik where to this day still produce particles of the highest class. When the production of particles is classified under the chemical process industry, the manufacturing process consists of a number of raw materials mixed and reacting with each other in different vessels. Many of the chemicals used are harmful to human health and therefore Nouryon is constantly working to increase safety while at the same time a great focus is on streamlining the manufacturing process. The purpose of the work is to reduce the variation in the manufacturing process by analyzing a particular product family, working with tools based on statistics and historical data, and automatic inspection systems with the basis in camera technology. By reducing the variation, the proportion of First Time Right will increase, which in turn results in reduced costs that arise due to rework. It would also facilitate production planning, reduce downtime for subsequent processes and generally result in a generally more reliable process. By raising the root causes of the variation and addressing these with the help of tools that simplify the daily work, control of the process can be taken and the needs of the customers be ensured. To create an understanding of the chemical process, a current study of how different manufacturing steps were carried out. In addition to this, it was also investigated how the company is currently working on raising problems to the surface and in order to be able to determine how the work with the statistical tools and the inspection system could be fullfilled in the current conditions. The work resulted in a large amount of data that was the basis for a multivariate analysis, a number of proposals for future work and also two separate reports discusses with statistical process control and inspection systems with camera technology. Unfortunately, no quantitative results can be presented in the work because all results are based on actions taken in the longer term.

This master thesis investigates the possibility to create a machine vision solution for the automation of earth-moving machines. This research was done as without some type of vision system it will not be possible to create a fully autonomous earth moving machine that can safely be used around humans or other machines. Cameras were used as the primary sensors as they are cheap, provide high resolution and is the type of sensor that most closely mimic the human vision system. The purpose of this master thesis was to use existing real time object detectors together with transfer learning and examine if they can successfully be used to extract information in environments such as construction, forestry and mining. The amount of data needed to successfully train a real time object detector was also investigated. Furthermore, the thesis examines if there are specifically difficult situations for the defined object detector, how reliable the object detector is and finally how to use service-oriented architecture principles can be used to create deep learning systems. To investigate the questions formulated above, three data sets were created where different properties were varied. These properties were light conditions, ground material and dump truck orientation. The data sets were created using a toy dump truck together with a similarly sized wheel loader with a camera mounted on the roof of its cab. The first data set contained only indoor images where the dump truck was placed in different orientations but neither the light nor the ground material changed. The second data set contained images were the light source was kept constant, but the dump truck orientation and ground materials changed. The last data set contained images where all property were varied. The real time object detector YOLOv3 was used to examine how a real time object detector would perform depending on which one of the three data sets it was trained using. No matter the data set, it was possible to train a model to perform real time object detection. Using a Nvidia 980 TI the inference time of the model was around 22 ms, which is more than enough to be able to classify videos running at 30 fps. All three data sets converged to a training loss of around 0.10. The data set which contained more varied data, such as the data set where all properties were changed, performed considerably better reaching a validation loss of 0.164 compared to the indoor data set, containing the least varied data, only reached a validation loss of 0.257. The size of the data set was also a factor in the performance, however it was not as important as having varied data. The result also showed that all three data sets could reach a mAP score of around 0.98 using transfer learning.

Tungsten Inert Gas (TIG) welding is extensively used in aerospace applications, due to its unique ability to produce higher quality welds compared to other shielded arc welding types. However, most TIG welding is performed manually and has not achieved the levels of automation that other welding techniques have. This is mostly attributed to the lack of process knowledge and adaptability to complexities, such as mismatches due to part fit-up. Recent advances in automation have enabled the use of industrial robots for complex tasks that require intelligent decision making, predominantly through sensors. Applications such as TIG welding of aerospace components require tight tolerances and need intelligent decision making capability to accommodate any unexpected variation and to carry out welding of complex geometries. Such decision making procedures must be based on the feedback about the weld profile geometry. In this thesis, a real-time position based closed loop system was developed with a six axis industrial robot (KUKA KR 16) and a laser triangulation based sensor (Micro-Epsilon Scan control 2900-25).

No matter how inspection techniques have been advanced, manual visual inspection is currently still the first and fundamental step in assessing civil infrastructure. If no sign of deterioration has been spotted in manual inspection, any future inspection actions is not necessary to take. However, manual inspection has been identified with several limitations including the qualitative nature of inspection results, the time-consuming inspection process, and the heavy reliance on inspectors' and/or engineers' experience. In order to overcome these limitations, automated visual inspection systems have been proposed to enhance and/or replicate the manual inspection process. A number of image processing methods have been developed in detecting defects (i.e. coating rusts) and damage (i.e. cracks) on civil infrastructure. Their effectiveness has been verified in inspecting structures, such as bridges, underground pipes, and tunnels. Although existing methods are effective in finding defects and damage from digital images, missing two important links limits their application for rapid infrastructure assessment and rehabilitation. The first link is the correlation between the defects/damage and the structural members that the defects/damage lie on. The second link is the relationship between the defects/damage and their impacts on the structural members. The purpose of this research is to investigate the way of establishing these two links. It is focused on the retrieval of critical structural members and defects/damage information from images/videos, and then the utilization of this information for automated and rapid assessment and rehabilitation of civil infrastructure. Specifically, a combination of techniques from the areas of visual pattern recognition, digital filtering, and machine vision have been used in order to develop a set of methods for concrete column recognition, crack properties retrieval, and air pockets and discoloration detection and evaluation. The methods proposed in this research were implemented in a Microsoft Visual Studio environment, and tested on the real images/videos of concrete structures inflicted with cracks, air pockets and discoloration. The test results indicated that the methods could automatically recognize concrete columns, correctly measure the properties of the cracks in a crack map, and accurately evaluate the impacts of air pockets and discoloration on the visual quality of concrete surfaces.

Στα πλαίσια της διατριβής μελετήθηκε αρχικά το πρόβλημα της εύρεσης της θέσης του υφάσματος επάνω στην τράπεζα εργασίας με μεθόδους ψηφιακής ανάλυσης σημάτων βίντεο, παρουσία φαινομένων παραμόρφωσης και μερικής επικάλυψης του υφάσματος. Οι νέες μέθοδοι εντοπισμού που αναπτύχθηκαν αξιολογήθηκαν πειραματικά παρουσιάζοντας ικανοποιητική ακρίβεια εντοπισμού και ανοχή του συστήματος σε μερικές επικαλύψεις και παραμορφώσεις. Μετά την ολοκλήρωση της ραφής του υφάσματος πραγματοποιείται αυτόματος εντοπισμός της θέσης της ραφής από ψηφιακές φωτογραφίες. Αναπτύχθηκαν τρείς πρωτότυπες μέθοδοι εντοπισμού της θέσης της ραφής οι οποίες διαφοροποιούνται στο στάδιο της προεπεξεργασίας. Η πειραματική αξιολόγηση γίνεται σε βάση δεδομένων που περιέχει 118 εικόνες έτοιμων ενδυμάτων. Προτού πραγματοποιηθεί ποιοτικός έλεγχος ραφής, οι εικόνες κανονικοποιούνται ως προς τη θέση και τον προσανατολισμό της ραφής χρησιμοποιώντας τις παραπάνω μεθόδους αυτόματου εντοπισμού της θέσης της ραφής. Αναπτύχθηκαν και αξιολογήθηκαν τρείς διαφορετικές μέθοδοι αυτόματης αναγνώρισης της ποιότητας σε δείγματα ραφής οι οποίες εξάγουν τρία διαφορετικά σύνολα χαρακτηριστικών. Η πρώτη μέθοδος βασίζεται σε φασματικά χαρακτηριστικά, η δεύτερη στην επιβολή αυτό-σκίασης, ενώ η τρίτη βασίζεται στην εκτίμηση της ανομοιομορφίας της επιφάνειας των δειγμάτων ραφής. Η πειραματική αξιολόγηση γίνεται σε βάση δεδομένων δειγμάτων ραφής που περιλαμβάνει 325 ραφές. Η εκτίμηση της ποιότητας ραφής πραγματοποιείται με ταξινόμηση σε πέντε διατεταγμένους βαθμούς ποιότητας. Σε αυτήν την κατεύθυνση, προτείνονται και συγκρίνονται τέσσερις μέθοδοι αναγνώρισης προτύπων διατεταγμένων κατηγοριών. Η πρώτη μέθοδος χρησιμοποιεί για την ταξινόμηση μοντέλο σύμμετρων αναλογιών πιθανότητας. Η δεύτερη μέθοδος κάνει αναγνώριση με χρήση γραμμικού μοντέλου. Οι άλλες δύο μέθοδοι είναι πρωτότυπες και επίσης χρησιμοποιούν γραμμικό μοντέλο για την ταξινόμηση. Η διαφοροποίησή τους από τη δεύτερη μέθοδο είναι ότι η επιλογή των αριθμητικών τιμών των διατεταγμένων κατηγορίων δεν γίνεται αυθαίρετα., αλλά προκύπτει ως λύση προβλημάτων ελαχιστοποίησης.. Η πειραματική αξιολόγηση και σύγκριση των μεθόδων στο πρόβλημα του ποιοτικού ελέγχου ραφών οδηγεί στην επιλογή του μοντέλου σύμμετρων αναλογιών πιθανότητας σε περίπτωση που υπάρχει ικανός αριθμός παραδειγμάτων εκπαίδευσης, ενώ σε αντίθετη περίπτωση μπορεί να προτιμηθεί το γραμμικό μοντέλο αφού προηγηθεί βελτιστοποίηση με χρήση κάποιας εκ των δύο προτεινόμενων μεθόδων επιλογής αριθμητικών τιμών. / One of the problems studied in the present dissertation is that of the detection of the fabrics’ position on the working area. The proposed detection methods are based on image processing and analysis techniques and take into consideration both partial occlusion and fabric deformation. The methods have been experimentally evaluated and the results indicate sufficient detection accuracy and robustness regarding partial occlusion and fabric deformation. After sewing the fabric, the position and orientation of the seam is automatically detected. Three novel seam detection methods have been developed using different pre-processing techniques. The experimental evaluation of the three detection methods is made on a database containing 118 images of ready sewn garments. Before performing seam quality control the seam images are normalized with respect to the seam position and orientation, using the aforementioned seam detection methods. Feature selection has been studied next, extracting three different sets of features and assessing seam quality using three different methods. The first method uses spectral features; the second method is based on the detection of self-shadows onto the seam specimens, while the third method is based on the estimation of the surface roughness of the specimens. The experimental evaluation of the proposed methods is made on a database containing 325 images of seam specimens. Seam quality control is performed by classifying the seam specimens into five ordinal grades of quality. In this direction, four classification methods are proposed and evaluated, taking into account the ordered arrangement of the classes. The first method uses the proportional odds model; while the second method uses a linear model. The other two methods are novel and also employ a linear model. The difference between these two methods and the second method is that the numerical values they are assigning to the ordered categories are not arbitrary like in the case of the second method. The experimental evaluation of these four methods indicates that in case of a large number of training data, the first method which is based on the proportional odds model is more efficient, while in case of an insufficient number of training data the linear model optimized by one of the two novel methods should be selected.

No description available.

Objekto atpažinimas ir pozicijos nustatymas gali būti pritaikomas daugeliui pramonėje egzistuojančių uždavinių. Šio darbo pagrindinis tikslas yra sukurti mašininės regos sistemą, kuria būtų galima greitai ir tiksliai rasti objekto poziciją pagal pasirinktą objekto modelį. Šiame darbe gilinamasi į GPU veikimo principus ir privalumus apdorojant vaizdus GLSL programavimo kalba. Apžvelgiami praktikoje taikomų metodų, skirtų objekto pozicijai nustatyti, veikimo principai, jų privalumai ir trūkumai. Taip pat šiame darbe aprašomas suformuotas ir įgyvendintas realaus laiko metodas, naudojantis GPU teikiama sparta atlikti vartotojo pasirinkto modelio paiešką. Pabaigoje pateikiami pasiekti įgyvendinto metodo spartos rodikliai, privalumai ir trūkumai. Darbą sudaro: įvadas, mašininėje regoje pasitaikančių problemų tyrinėjimas, objekto paieškos metodų apžvalga, darbo su grafinėmis vaizdo plokštėmis privalumai ir trūkumai, objekto paieškos su grafine vaizdo plokšte metodas, pasiekti rezultatai, išvados ir literatūros sąrašas. Darbo apimtis – 53 p. teksto be priedų, 30 pav., 2 lent., 26 literatūros šaltiniai. / Object recognition and parameter detection could be used in many areas from electronics to food industry. One of the most important problems in laser industry is to transform laser work trajectories based on constant object model. In the real life applications model could be rotated or translated due to the fact that object must be put in laser work area. The translation and rotation of object must be found to fit user defined constant model. There are many methods for object parameters detection, but image processing tasks require a lot of computing power. Recent research on image processing with graphics processing units - GPU, shows huge performance results compared with central processing units – CPU. The purpose of this work is to find out the main fundamentals for fast and accurate object parameter detection in machine vision systems. In this work it is focused on object parameter detection with GPU. Moreover, the analysis and comparison of different object parameters detection methods are proposed. Object parameter detection system was implemented with C++ and GLSL shading language, thus the system could be adapted to different computer hardware and operating systems. Work size – 53 p. text, 30 illustrations, 2 tables, 26 bibliographic sources.

Orientadores: Pedro Jussieu de Rezende, Alexandre Xavier Falcão / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Computação / Made available in DSpace on 2018-08-18T13:40:27Z (GMT). No. of bitstreams: 1 CasteloFernandez_CesarChristian_M.pdf: 2721705 bytes, checksum: 0d764319e69f64e1b806f60bbbf54b92 (MD5) Previous issue date: 2011 / Resumo: O Reconhecimento de Padrões pode ser definido como a capacidade de identificar a classe de algum objeto dentre um dado conjunto de classes, baseando-se na informação fornecida por amostras conhecidas (conjunto de treinamento). Nesta dissertação, o foco de estudo é o paradigma de classificação supervisionada, no qual se conhece a classe de todas as amostras utilizadas para o projeto do classificador. Especificamente, estuda-se o Classificador baseado em Floresta de Caminhos Ótimos (Optimum-Path Forest - OPF) e propõem três novos algoritmos de aprendizado, os quais representam melhorias em comparação com o Classificador OPF tradicional. Primeiramente, é desenvolvida uma metodologia simples, porém efetiva, para detecção de outliers no conjunto de treinamento. O método visa uma melhoria na acurácia do Classificador OPF tradicional através da troca desses outliers por novas amostras do conjunto de avaliação e sua exclusão do processo de aprendizagem. Os outliers são detectados computando uma penalidade para cada amostra baseada nos seus acertos e erros na classificação, o qual pode ser medido através do número de falsos positivos/negativos e verdadeiros positivos/negativos obtidos por cada amostra. O método obteve uma melhoria na acurácia em comparação com o OPF tradicional, com apenas um pequeno aumento no tempo de treinamento. Em seguida, é proposto um aprimoramento ao primeiro algoritmo, que permite detectar com maior precisão os outliers presentes na base de dados. Neste caso, utiliza-se a informação de falsos positivos/negativos e verdadeiros positivos/negativos de cada amostra para explorar intrinsecamente as relações de adjacência de cada amostra e determinar se é outlier. Uma inovação do método é que não existe necessidade de se computar explicitamente tal adjacência, como é feito nas técnicas tradicionais, o qual pode ser inviável para grandes bases de dados. O método obteve uma boa taxa de detecção de outliers e um tempo de treinamento muito baixo em vista do tamanho das bases de dados utilizadas. Finalmente, é abordado o problema de se selecionar um úmero tão pequeno quanto possível de amostras de treinamento e se obter a maior acurácia possível sobre o conjunto de teste. Propõe-se uma metodologia que se inicia com um pequeno conjunto de treinamento e, através da classificação de um conjunto bem maior de avaliação, aprende quais amostras são as mais representativas para o conjunto de treinamento. Os resultados mostram que é possível obter uma melhor acurácia que o Classificador OPF tradicional ao custo de um pequeno incremento no tempo de treinamento, mantendo, no entanto, o conjunto de treinamento menor que o conjunto inicial, o que significa um tempo de teste reduzido / Abstract: Pattern recognition can be defined as the capacity of identifying the class of an object among a given set of classes, based on the information provided by known samples (training set). In this dissertation, the focus is on the supervised classification approach, for which we are given the classes of all the samples used in the design of the classifier. Specifically, the Optimum-Path Forest Classifier (OPF) is studied and three new learning algorithms are proposed, which represent improvements to the traditional OPF classifier. First of all, a simple yet effective methodology is developed for the detection of outliers in a training set. This method aims at improving OPF's accuracy through the swapping of outliers for new samples from the evaluating set and their exclusion from the learning process itself. Outliers are detected by computing a penalty for each sample based on its classification-hits and -misses, which can be measured through the number of false positive/negatives and true positives/negatives obtained by each sample. The method achieved an accuracy improvement over the traditional OPF, with just a slight increment in the training time. An improvement to the first algorithm is proposed, allowing for a more precise detection of outliers present in the dataset. In this case, the information on the number of false positive/negatives and true positives/negatives of each sample is used to explore the adjacency relations of each sample and determine whether it is an outlier. The method's merit is that there is no need of explicitly computing an actual vicinity, as the traditional techniques do, which could be infeasible for large datasets. The method achieves a good outlier detection rate and a very low training time, considering the size of the datasets. Finally, the problem of choosing a small number of training samples while achieving a high accuracy in the testing set is addressed. We propose a methodology which starts with a small training set and, through the classification of a much larger evaluating set, it learns which are the most representative samples for the training set. The results show that it is possible to achieve higher accuracy than the traditional OPF's at the cost of a slight increment in the training time, preserving, however, a smaller training set than the original one, leading to a lower testing time / Mestrado / Ciência da Computação / Mestre em Ciência da Computação

Samarbetspartnern som denna studie har gjorts för har som mål att konstruera en detektor för tågtrafiken som bygger på bildigenkänning och artificiell intelligens. Problemet är att de lösningar som finns idag är dyra, och därför är en förutsättning att den ska vara byggd med konsumentprodukter för att få ner kostnaden samt att den ska vara enkel att installera och underhålla. Flera ramverk för objektigenkänning existerar, men dessa bygger på olika metoder och tekniker. Studien har därför utförts som en fallstudie vars syfte har varit att jämföra två välanvända ramverk för objektigenkänning för att identifiera olika för- och nackdelar gällande noggrannhet och tidsåtgång vid träning och test med hjälp av dessa ramverk. Även vilka olika utmaningar som stötts på under tillvägagångssättet har lyfts fram. Studien sammanfattar sedan dessa för att skapa idéer och diskussion för hur dessa skulle kunna implementeras på den nya tågdetektorn. Ramverken som har jämförts är OpenCV och Google TensorFlow. Dessa bygger på olika objektigenkänningstekniker, i huvudsak kaskadklassificering och neurala nät. Ramverken testades med en datamängd på 400 bilder på olika tågfordon där hjulaxlarna användes som parameter för objektigenkänningen. Testerna bedömdes efter kriterier gällande noggrannhet, tidsåtgång för träning samt komplexitet för konfiguration och användning. Resultatet visade att OpenCV hade en snabb träningsprocess, men visade låg precision och en mer komplex konfigurerings- och användningsprocess. TensorFlow hade en långsammare träningsprocess, men visade istället bättre precision och en mindre komplex konfigurering. Slutsatsen av studien är att TensorFlow visade bäst resultat och har mest potential att användas i den nya tågdetektorn. Detta baseras på studiens resultat samt att den bygger på modernare tekniker med neurala nät för objektigenkänning. / The research in this thesis is conducted with the partners aim to construct a new train detection system that uses image recognition and artificial intelligence. Detectors like these that exists today are expensive, so the construction is going to be based around the use of consumer electronics to lower the cost and simplify installation and maintenance. Several frameworks for object detection are available, but they use different approaches and methods. This thesis is therefore carried out as a case study that compares two widely used frameworks for image recognition tasks. The purpose is to identify advantages and disadvantages regarding training and testing when using these frameworks. Also highlighted is different challenges encountered in the process. The summary of the results is used to form ideas and a discussion about how to implement a framework in the new detection system. The frameworks compared in this study are OpenCV and Google TensorFlow. These frameworks use different methods for object detection, mainly cascade classifiers and convolutional neural nets. The frameworks were tested using a dataset of 400 images on different trains where the wheel-axles were used as the object of interest. The results were analyzed based on criteria regarding precision, total training time and also complexity regarding configuration and usage. The results showed that OpenCV had a faster training process but had low precision and more complex configuration. TensorFlow had a much longer training process but had better precision and less complex configuration. The conclusion of the study is that TensorFlow overall showed the best result and has a better potential for implementation in the new detection system. This is based on the results from the study, but also that the framework is developed with a more modern approach using convolutional neural nets for bject detection.

The diploma thesis deals with the usage of integrated machine vision by B&R, Smart Sensor, for metal rectangular-shaped workpiece recognition and position determinition. The description of the usage of machine vision in the industry is made, the solution concept is created and the program and the user visualisation implemented. The main outcome of this work is the VITemplate library which can be used in combination with the model-based Blob analysis implemented in Smart Sensor to control the robotic arm to successfully grab all the workpieces on the belt.