Jens Brandenburger, Fabian Krippendorff, Michael Krätzner, Michael Nörtersheuser, Xin Chen, A. Boss, K. Jonker, Nicolas Pipard, A. Ebner
Zusammenfassung In der modernen Stahlproduktion sind automatische Oberflächeninspektionssysteme (OIS) zur Detektion und Klassifikation von Oberflächenfehlern weit verbreitet und ihre Ergebnisse haben stark an Bedeutung gewonnen. Trotzdem fehlt es bis heute an anerkannten Methoden für eine objektive und umfassende Leistungsbewertung der Systeme, um mit vertretbarem Aufwand geeignete Kenngrößen für die OIS-Klassifikationsleistung im realen Produktionsbetrieb zu ermitteln. Dieser Beitrag widmet sich der Problematik der Abschätzung des Recalls bei unbekannter „Grundwahrheit“ (ground truth), als zentrales Maß für die Fähigkeitsbewertung klassifizierender Bildverarbeitungssysteme (BV-Systeme). Es werden eine Methodik für die Recall-Schätzung mittels Hilfsklassifikator vorgestellt und Forschungsbedarfe für deren praktische Umsetzung erörtert.
{"title":"Quantifizierung der Klassifikationsleistung von Oberflächeninspektionssystemen in der Flachstahlproduktion","authors":"Jens Brandenburger, Fabian Krippendorff, Michael Krätzner, Michael Nörtersheuser, Xin Chen, A. Boss, K. Jonker, Nicolas Pipard, A. Ebner","doi":"10.1515/teme-2023-0035","DOIUrl":"https://doi.org/10.1515/teme-2023-0035","url":null,"abstract":"Zusammenfassung In der modernen Stahlproduktion sind automatische Oberflächeninspektionssysteme (OIS) zur Detektion und Klassifikation von Oberflächenfehlern weit verbreitet und ihre Ergebnisse haben stark an Bedeutung gewonnen. Trotzdem fehlt es bis heute an anerkannten Methoden für eine objektive und umfassende Leistungsbewertung der Systeme, um mit vertretbarem Aufwand geeignete Kenngrößen für die OIS-Klassifikationsleistung im realen Produktionsbetrieb zu ermitteln. Dieser Beitrag widmet sich der Problematik der Abschätzung des Recalls bei unbekannter „Grundwahrheit“ (ground truth), als zentrales Maß für die Fähigkeitsbewertung klassifizierender Bildverarbeitungssysteme (BV-Systeme). Es werden eine Methodik für die Recall-Schätzung mittels Hilfsklassifikator vorgestellt und Forschungsbedarfe für deren praktische Umsetzung erörtert.","PeriodicalId":56086,"journal":{"name":"Tm-Technisches Messen","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-06-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73798110","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
For many technical applications, obtaining sensory information about objects, a scene or the environment is crucial. These include, for example, determining product quality in quality assurance and sensor-based sorting, sensing the environment for robotics and automated vehicles, and many other tasks in measurement and automation technology. In all of these applications, machine vision systems have key advantages over other sensor principles and over the inspection by humans: The actual observation process—image acquisition—is contact-free, the data have a high information content due to their multi-dimensional nature, and a variety of image acquisition methods can be used to capture very different properties of the scene with high informative value. What is outstanding about machine vision, however, is that it emulates the most important human sense—the visual sense—so that many image processing procedures can be understood relatively easily by humans. On the other hand, technical image acquisition is not bound to the limitations of the human sense of sight (e. g., spectral sensitivity, temporal response, temporal and spatial resolution, reproducibility, objectivity, fatigue). Cameras and the images they capture also play an increasing role in daily life, which is immediately apparent from the omnipresence of smartphones with (now often multiple) cameras. This is accompanied by a high level of maturity in sensor technology and image data processing, which in turn benefits the technical applications of image processing. In machine vision systems, components of various disciplines, including optics, lighting technology, sensor technology, signal processing, system theory, computer science and information technology, interact with each other to
{"title":"Image Processing Forum – Forum Bildverarbeitung 2022","authors":"Thomas Längle, M. Heizmann","doi":"10.1515/teme-2023-0093","DOIUrl":"https://doi.org/10.1515/teme-2023-0093","url":null,"abstract":"For many technical applications, obtaining sensory information about objects, a scene or the environment is crucial. These include, for example, determining product quality in quality assurance and sensor-based sorting, sensing the environment for robotics and automated vehicles, and many other tasks in measurement and automation technology. In all of these applications, machine vision systems have key advantages over other sensor principles and over the inspection by humans: The actual observation process—image acquisition—is contact-free, the data have a high information content due to their multi-dimensional nature, and a variety of image acquisition methods can be used to capture very different properties of the scene with high informative value. What is outstanding about machine vision, however, is that it emulates the most important human sense—the visual sense—so that many image processing procedures can be understood relatively easily by humans. On the other hand, technical image acquisition is not bound to the limitations of the human sense of sight (e. g., spectral sensitivity, temporal response, temporal and spatial resolution, reproducibility, objectivity, fatigue). Cameras and the images they capture also play an increasing role in daily life, which is immediately apparent from the omnipresence of smartphones with (now often multiple) cameras. This is accompanied by a high level of maturity in sensor technology and image data processing, which in turn benefits the technical applications of image processing. In machine vision systems, components of various disciplines, including optics, lighting technology, sensor technology, signal processing, system theory, computer science and information technology, interact with each other to","PeriodicalId":56086,"journal":{"name":"Tm-Technisches Messen","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-06-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88512389","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
P. Wilhelm, M. Eggert, Stefan Oertel, Julia Hornig
Zusammenfassung Das an der Physikalisch-Technischen Bundesanstalt (PTB) entwickelte Wind-Lidar misst Windgeschwindigkeit, Windrichtung und Messhöhe mit hoher zeitlicher und örtlicher Auflösung sowie geringer Messunsicherheit. Das mobile PTB-Wind-Lidar ermöglicht hochgenaue und auf die SI-Einheiten rückführbare optische Windfernmessungen in Höhen zwischen 5 m und 250 m, wie sie in der Windindustrie und Meteorologie benötigt werden. Da das System keine geländeabhängigen Korrekturfaktoren benötigt, ermöglicht es präzise und hochaufgelöste Messungen auch vor und im Nachlauf von Windenergieanlagen. In diesem Artikel werden der Aufbau und die Funktionsweise des Messsystems beschrieben, einschließlich der bistatischen Geometrie, des faseroptischen Aufbaus, der Signalverarbeitung und der Messvolumengeometrie. Die hohe Auflösung des Messsystems wird anhand von ausgewählten Datensätzen erstmals in Form von Spektrogrammen verdeutlicht. Eine Zusammenstellung zuvor publizierter Vergleichsmessungen zeigt die Leistungsfähigkeit des PTB-Wind-Lidars auf. Die Vergleichsinstrumente umfassen ein Ultraschallanemometer, ein Laser-Doppler-Anemometer sowie ein Schalensternanemometer.
{"title":"Ein mobiles System zur vektoriellen Messung der Windgeschwindigkeit","authors":"P. Wilhelm, M. Eggert, Stefan Oertel, Julia Hornig","doi":"10.1515/teme-2023-0055","DOIUrl":"https://doi.org/10.1515/teme-2023-0055","url":null,"abstract":"Zusammenfassung Das an der Physikalisch-Technischen Bundesanstalt (PTB) entwickelte Wind-Lidar misst Windgeschwindigkeit, Windrichtung und Messhöhe mit hoher zeitlicher und örtlicher Auflösung sowie geringer Messunsicherheit. Das mobile PTB-Wind-Lidar ermöglicht hochgenaue und auf die SI-Einheiten rückführbare optische Windfernmessungen in Höhen zwischen 5 m und 250 m, wie sie in der Windindustrie und Meteorologie benötigt werden. Da das System keine geländeabhängigen Korrekturfaktoren benötigt, ermöglicht es präzise und hochaufgelöste Messungen auch vor und im Nachlauf von Windenergieanlagen. In diesem Artikel werden der Aufbau und die Funktionsweise des Messsystems beschrieben, einschließlich der bistatischen Geometrie, des faseroptischen Aufbaus, der Signalverarbeitung und der Messvolumengeometrie. Die hohe Auflösung des Messsystems wird anhand von ausgewählten Datensätzen erstmals in Form von Spektrogrammen verdeutlicht. Eine Zusammenstellung zuvor publizierter Vergleichsmessungen zeigt die Leistungsfähigkeit des PTB-Wind-Lidars auf. Die Vergleichsinstrumente umfassen ein Ultraschallanemometer, ein Laser-Doppler-Anemometer sowie ein Schalensternanemometer.","PeriodicalId":56086,"journal":{"name":"Tm-Technisches Messen","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90465795","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract This contribution proposes an approach and the respective tool that uses Active Deep Learning (ADL) to segment industrial three-dimensional computed tomography (3D CT) data. The general approach is application-independent and includes an iterative human-in-the-loop Active Learning (AL) process that produces labeled training data and a trained Deep Learning (DL) model for semantic segmentation. The model is continuously improved during iterations such that manual labeling effort is reduced. In addition, the user can minimize user interaction with the aid of a random forest-based classifier and focus on unclear or invalid segmentation results. The complete workflow is implemented within one single Python tool. The approach is demonstrated in detail for two industrial use cases: Single fiber analysis and plant segmentation. For plant segmentation, the method is compared to a baseline and a classic image processing algorithm.
{"title":"Active deep learning for segmentation of industrial CT data","authors":"Markus Michen, M. Rehak, U. Hassler","doi":"10.1515/teme-2023-0047","DOIUrl":"https://doi.org/10.1515/teme-2023-0047","url":null,"abstract":"Abstract This contribution proposes an approach and the respective tool that uses Active Deep Learning (ADL) to segment industrial three-dimensional computed tomography (3D CT) data. The general approach is application-independent and includes an iterative human-in-the-loop Active Learning (AL) process that produces labeled training data and a trained Deep Learning (DL) model for semantic segmentation. The model is continuously improved during iterations such that manual labeling effort is reduced. In addition, the user can minimize user interaction with the aid of a random forest-based classifier and focus on unclear or invalid segmentation results. The complete workflow is implemented within one single Python tool. The approach is demonstrated in detail for two industrial use cases: Single fiber analysis and plant segmentation. For plant segmentation, the method is compared to a baseline and a classic image processing algorithm.","PeriodicalId":56086,"journal":{"name":"Tm-Technisches Messen","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83569244","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Timo Zander, Ziyan Pan, Pascal Birnstill, J. Beyerer
Abstract Anomaly detection with machine learning in industrial inspection systems for manufactured products relies on labelled data. This raises the question of how the labelling by humans should be conducted. Moreover, such a system will most likely always be imperfect and potentially need a human fall-back mechanism for ambiguous cases. We consider the case where we want to optimise the cost of the combined inspection process done by humans together with a pre-trained algorithm. This gives improved combined performance and increases the knowledge of the performance of the pre-trained model. We focus on so-called one-class classification problems which produce a continuous outlier score. After establishing some initial setup mechanisms ranging from using prior knowledge to calibrated models, we then define some cost model for machine inspection with a possible second inspection of the sample done by a human. Further, we discuss in this cost model how to select two optimal boundaries of the outlier score, where in between these two boundaries human inspection takes place. Finally, we frame this established knowledge into an applicable algorithm and conduct some experiments for the validity of the model.
{"title":"Finding optimal decision boundaries for human intervention in one-class machine-learning models for industrial inspection","authors":"Timo Zander, Ziyan Pan, Pascal Birnstill, J. Beyerer","doi":"10.1515/teme-2023-0010","DOIUrl":"https://doi.org/10.1515/teme-2023-0010","url":null,"abstract":"Abstract Anomaly detection with machine learning in industrial inspection systems for manufactured products relies on labelled data. This raises the question of how the labelling by humans should be conducted. Moreover, such a system will most likely always be imperfect and potentially need a human fall-back mechanism for ambiguous cases. We consider the case where we want to optimise the cost of the combined inspection process done by humans together with a pre-trained algorithm. This gives improved combined performance and increases the knowledge of the performance of the pre-trained model. We focus on so-called one-class classification problems which produce a continuous outlier score. After establishing some initial setup mechanisms ranging from using prior knowledge to calibrated models, we then define some cost model for machine inspection with a possible second inspection of the sample done by a human. Further, we discuss in this cost model how to select two optimal boundaries of the outlier score, where in between these two boundaries human inspection takes place. Finally, we frame this established knowledge into an applicable algorithm and conduct some experiments for the validity of the model.","PeriodicalId":56086,"journal":{"name":"Tm-Technisches Messen","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86714617","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract We have developed a novel metric to gauge the performance of artificial intelligence (AI) or machine learning (ML) algorithms, called the Spatial Recall Index (SRI). The novelty is the spatial resolution of a standard performance indicator, as a Recall value is assigned to each individual pixel. This generates a distribution of the performance of a given AI-algorithm with the resolution of the images in the dataset. While the mathematical basis has already been presented before, here we demonstrate the usage on more datasets and delve into in-depth application examples. We examine both the MS COCO and the Berkeley Deep Drive datasets, using a state-of-the-art object detection algorithm. The dataset is degraded using a physical-realistic lens-model, where the optical performance varies over the field of view, as a real camera would. This study highlights the usefulness of the SRI, as every image has been taken by realistic optics. A generalization, the GSRI is introduced, from which we derive SRIA, weighting with object area and SRIrisk intended for autonomous driving. Finally, these metrics are compared.
{"title":"Local performance evaluation of AI-algorithms with the generalized spatial recall index","authors":"Patrick Müller, Alexander Braun","doi":"10.1515/teme-2023-0013","DOIUrl":"https://doi.org/10.1515/teme-2023-0013","url":null,"abstract":"Abstract We have developed a novel metric to gauge the performance of artificial intelligence (AI) or machine learning (ML) algorithms, called the Spatial Recall Index (SRI). The novelty is the spatial resolution of a standard performance indicator, as a Recall value is assigned to each individual pixel. This generates a distribution of the performance of a given AI-algorithm with the resolution of the images in the dataset. While the mathematical basis has already been presented before, here we demonstrate the usage on more datasets and delve into in-depth application examples. We examine both the MS COCO and the Berkeley Deep Drive datasets, using a state-of-the-art object detection algorithm. The dataset is degraded using a physical-realistic lens-model, where the optical performance varies over the field of view, as a real camera would. This study highlights the usefulness of the SRI, as every image has been taken by realistic optics. A generalization, the GSRI is introduced, from which we derive SRIA, weighting with object area and SRIrisk intended for autonomous driving. Finally, these metrics are compared.","PeriodicalId":56086,"journal":{"name":"Tm-Technisches Messen","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83474915","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Themenheft Sensoren und Messsystem 2022","authors":"J. Wöllenstein, L. Reindl","doi":"10.1515/teme-2023-0092","DOIUrl":"https://doi.org/10.1515/teme-2023-0092","url":null,"abstract":"","PeriodicalId":56086,"journal":{"name":"Tm-Technisches Messen","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81929241","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract An almost lossless compression method for images is introduced adapted to the temporal noise of image sensors. In a first step, a non-linear gray value transform is applied to generate an image with a gray value independent temporal noise and less bits than the original image. The chosen value for the standard deviation of the temporal noise in the transformed image determines how accurately mean values and the standard deviation of temporal noise can be computed and to which extent the image can be compressed further by a lossless compression in a second step. Just a measurement of the noise characteristics according to the open and international EMVA standard 1288, a non-linear gray value transform for noise equalization, and an open source lossless compression algorithm are required to use this new compression method.
{"title":"Almost lossless compression of noisy images","authors":"B. Jähne","doi":"10.1515/teme-2023-0028","DOIUrl":"https://doi.org/10.1515/teme-2023-0028","url":null,"abstract":"Abstract An almost lossless compression method for images is introduced adapted to the temporal noise of image sensors. In a first step, a non-linear gray value transform is applied to generate an image with a gray value independent temporal noise and less bits than the original image. The chosen value for the standard deviation of the temporal noise in the transformed image determines how accurately mean values and the standard deviation of temporal noise can be computed and to which extent the image can be compressed further by a lossless compression in a second step. Just a measurement of the noise characteristics according to the open and international EMVA standard 1288, a non-linear gray value transform for noise equalization, and an open source lossless compression algorithm are required to use this new compression method.","PeriodicalId":56086,"journal":{"name":"Tm-Technisches Messen","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76506114","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Zusammenfassung In der PTB ist innerhalb der letzten Dekade umfangreiche Forschung zur Entwicklung einer zuverlässigen Messtechnik für Großverzahnungen betrieben worden. Die Arbeiten waren wesentlich für den Aufbau eines Kompetenzzentrums WIND, das im Jahr 2021 in den Wirkbetrieb gegangen ist. Mit dem Ziel, über eine weltweit einzigartige metrologische Infrastruktur Kalibrierdienstleistungen für Verzahnungen mit Durchmessern von einigen Metern anbieten zu können, sind systematisch die für eine aufgabenspezifische Rückführung erforderlichen Aspekte untersucht und entwickelt worden. In diesem Übersichtsartikel werden die wichtigsten Ergebnisse vorgestellt. Ein Großverzahnungsnormal mit einem Durchmesser von 2 m und einer Masse von fast 3 t konnte mit Unsicherheiten unterhalb von 3,5 µm (k = 2) kalibriert werden und dient seither als Referenz für alle weiteren Untersuchungen. Dazu gehört die Charakterisierung der wichtigsten Unsicherheitseinflüsse wie Temperatur, Messgeräteabweichungen und Werkstückmasseneffekte.
{"title":"Verzahnungsmetrologie für Windenergieanlagen","authors":"M. Stein, K. Kniel","doi":"10.1515/teme-2023-0053","DOIUrl":"https://doi.org/10.1515/teme-2023-0053","url":null,"abstract":"Zusammenfassung In der PTB ist innerhalb der letzten Dekade umfangreiche Forschung zur Entwicklung einer zuverlässigen Messtechnik für Großverzahnungen betrieben worden. Die Arbeiten waren wesentlich für den Aufbau eines Kompetenzzentrums WIND, das im Jahr 2021 in den Wirkbetrieb gegangen ist. Mit dem Ziel, über eine weltweit einzigartige metrologische Infrastruktur Kalibrierdienstleistungen für Verzahnungen mit Durchmessern von einigen Metern anbieten zu können, sind systematisch die für eine aufgabenspezifische Rückführung erforderlichen Aspekte untersucht und entwickelt worden. In diesem Übersichtsartikel werden die wichtigsten Ergebnisse vorgestellt. Ein Großverzahnungsnormal mit einem Durchmesser von 2 m und einer Masse von fast 3 t konnte mit Unsicherheiten unterhalb von 3,5 µm (k = 2) kalibriert werden und dient seither als Referenz für alle weiteren Untersuchungen. Dazu gehört die Charakterisierung der wichtigsten Unsicherheitseinflüsse wie Temperatur, Messgeräteabweichungen und Werkstückmasseneffekte.","PeriodicalId":56086,"journal":{"name":"Tm-Technisches Messen","volume":null,"pages":null},"PeriodicalIF":1.0,"publicationDate":"2023-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86471433","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：