Computer Vision for X-Ray Testing: Imaging, Systems, Image Databases, and Algorithms PDF
Preview Computer Vision for X-Ray Testing: Imaging, Systems, Image Databases, and Algorithms
Domingo Mery Computer Vision for X-Ray Testing Imaging, Systems, Image Databases, and Algorithms Computer Vision for X-Ray Testing Domingo Mery Computer Vision for X-Ray Testing Imaging, Systems, Image Databases, and Algorithms 123 DomingoMery Pontificia Universidad Católica deChile Santiago Chile ISBN978-3-319-20746-9 ISBN978-3-319-20747-6 (eBook) DOI 10.1007/978-3-319-20747-6 LibraryofCongressControlNumber:2015942783 SpringerChamHeidelbergNewYorkDordrechtLondon ©SpringerInternationalPublishingSwitzerland2015 Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpart of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission orinformationstorageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilar methodologynowknownorhereafterdeveloped. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publicationdoesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfrom therelevantprotectivelawsandregulationsandthereforefreeforgeneraluse. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authorsortheeditorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinor foranyerrorsoromissionsthatmayhavebeenmade. Printedonacid-freepaper SpringerInternationalPublishingAGSwitzerlandispartofSpringerScience+BusinessMedia (www.springer.com) To Ximena, Anais and Valeria who show me everyday the X-rays of love Foreword The wavelengths of X-rays are far shorter than those of visible light, and even shorter than those of ultraviolet light. Wilhelm Conrad Röntgen (1845–1923) was awarded the first Nobel prize in Physics in 1901 for his contributions to the detectionofelectromagneticradiation,andtothegenerationofX-rays,whicharea form of electromagnetic radiation. Radiographs are produced by having X-rays, emitted from a source, geometrically assumed to be a point in three-dimensional (3D)space,recordedonascreen.Thisscreenmighthaveaslightlycurvedsurface, but we can also see it (via defined mapping) as an image plane. X-ray technology provides a way to visualize the inside of visually opaque objects.Pixelintensitiesinrecordedradiographscorrespondbasicallytothedensity of matter, integrated along rays; those readers interested in a more accurate description may wish to look up the interaction of X-rays with matter by way of photo-absorption, Compton scattering, or Rayleigh scattering by reading the first chapter of this book. X-ray technology aims at minimizing scattering, by having nearly perfect rays passthroughthestudiedobject.Thus,wehaveaveryparticularimagingmodality: objects of study need to fit into a bounded space, defined as being between source and image plane, and pixel intensities have a meaning which differs from our commonly recorded digital images when using optical cameras. When modeling an X-ray imaging system we can apply much of the projective geometry, mathematics in homogeneous spaces, or analogous parameter notations: wejustneedtobeawarethatwearelooking“backwards,”fromtheimageplaneto the source (known as projection center), and no longer from the image plane into thepotentiallyinfinitespaceinfrontofanopticalcamera.Thus,itappearsthatthe problem of understanding 3D objects is greatly simplified by simply studying a bounded space: using a finite number of source-plus-screen devices for recording this bounded space; applying photogrammetric methods for understanding multi-view recordings, and applying the proper interpretation (e.g., basically den- sity) to the corresponding pixel values. Thus, this very much follows a common scenario of a computer vision, while also including image preprocessing and vii viii Foreword segmentation, object detection, and classification. The book addresses all of these subjects in the particular context of X-ray testing based on computer vision. The briefly sketched similarities between common (i.e., optical-camera-based) computer vision and X-ray testing techniques might be a good motif to generate curiosity among people working in computer vision, in order to understand how their knowledge can contribute to, or benefit from, various methods of X-ray testing. ThebookillustratesX-raytestingforaninterestingrangeofapplications.Italso introducesapublicallyavailablesoftwaresystemandanextensiveX-raydatabase. The book will undoubtedly contribute to the popularity of X-ray testing among thoseinthecomputervisionandimageanalysiscommunity,andmayalsoserveas a textbook or as support material for undertaking related research. Auckland Reinhard Klette April 2015 Preface Thisbookhasbeenwritteninmanyspatiotemporalcoordinates.Forinstance,some equationsandfigureswereperformedduringmyPh.D.attheTechnicalUniversity of Berlin (1996–2000). During that period, but in Hamburg, I took several X-ray images—that have been used in this book—in YXLON X-ray International Labs. After completing my Ph.D., and during my work in Santiago, Chile as associate researcherattheUniversity ofSantiagoofChile (2001–2003) andfacultymember at the Catholic University of Chile (2004–to date) I have written more than 40 journalpapersoncomputervisionappliedtoX-raytesting.Duringthistime,Ihave developedaMatlab Toolboxthathasbeenusedinmyresearchprojectsandinmy classes teaching image processing, pattern recognition and computer vision for graduateandundergraduatestudents.Overthelastfewyears,mygraduatestudents have taken thousands of X-ray images in our X-ray Testing Lab at the Catholic University of Chile. Moreover, in my sabbatical year at the University of Notre Dame (2014–2015), I had the time and space to teach the computer vision course for students of computer sciences, electrical engineering, and physics, and I have been able to bring together all those related papers, diagrams, and codes in this book. The present work has been written not only in three different countries (Germany, Chile, and the United States) over the last 15 years, but also in many different small places that provided me with the time and peace to write a para- graph,acaptionofafigure,acode,orwhateverIcould.Forexample,Iremembera Café in Michigan City where I spent various hours last winter writing this book withadeliciouscappuccinobesideme;ormystudyroominFisherApartmentson Notre Dame Campus, looking out the window at a squirrel holding a nut; or on a narrow tray table while taking an Inter-Regio train between Berlin and Hamburg, which was where I drew a diagram using a pen and probably a napkin; and of course,mydelightfulofficeattheCatholicUniversityofChilewithitsbreathtaking view of the Andes Mountains. Thisbookhasbeenputtogetheronthebasisoffourmainpillarsthathavebeen constructedoverthelast15years:thefirstpillaristhesetofjournalandconference papers that I have published. The second corresponds to the material used in my ix x Preface classes and the feedback received from students when I have been teaching image processing,patternrecognition,andcomputer vision.Thethirdpillar istheMatlab Toolbox that I was able to develop during this time, and which has been tested in severalexperiments,classes,andresearchprojects,amongothers.Thefourthpillar is the thousands of X-ray images that my research group has been taking in recent yearsatourLab,andtheX-rayimagesofdiecastingsthatItookinHamburg.Over allthistime,Ihaverealizedthatthisamountofworkcanallbebroughttogetherin a book that collects the most important contributions in computer vision used in X-ray testing. Scope X-ray imaging has been developed not only for its use in medical imaging for humans, but also for materials or objects, where the aim is to analyze—nonde- structively—those inner parts that are undetectable to the naked eye. Thus, X-ray testing is used to determine if a test object deviates from a given set of specifica- tions. Typical applications are analysis of food products, screening of baggage, inspection of automotive parts, and quality control of welds. In order to achieve efficient and effective X-ray testing, automated and semi-automated systems are beingdevelopedtoexecutethistask.Inthisbook,wepresentageneraloverviewof computer vision methodologies that have been used in X-ray testing. In addition, some techniques that have been applied in certain relevant applications are pre- sented: there are also some areas—like casting inspection—where automated sys- tems are very effective, and other application areas—such as baggage screening— where human inspection is still used. There are certain application areas—like welds and cargo inspections—where the process is semi-automatic; and there is someresearchinareas—includingfoodanalysis—whereprocessesarebeginningto be characterized by the use of X-ray imaging. In this book, Matlab programs for image analysis and computer vision algorithms are presented with real X-ray images that are available in a public database created for testing and evaluation. Organization The book is organized as follows: Chapter 1 (X-ray Testing): This chapter provides an introduction to the book. ItillustratesprinciplesaboutthephysicsofX-rays,anddescribesX-raytestingand imaging systems, while also summarizing the most important issues on computer vision for X-ray testing. Chapter2(ImagesforX-rayTesting):Thischapterpresentsadescriptionofthe GDXraydatabase,thedatasetofmorethan19,400X-rayimagesusedinthisbook Preface xi to illustrate and test several computer vision methods. The database includes five groups of X-ray images: castings, welds, baggage, natural objects and settings. Chapter 3 (Geometry in X-ray Testing): This chapter presents a mathematical backgroundofthemonocular andmultiple view geometrythat isnormallyusedin X-ray computer vision systems. Chapter4(X-rayImageProcessing):Thissectioncoversthemaintechniquesof image processing used in X-ray testing, such as image pre-processing, image fil- tering, edge detection, image segmentation, and image restoration. Chapter5(X-rayImageRepresentation):Thischaptercoversseveraltopicsthat areusedtorepresentanX-rayimage(oraspecificregionofanX-rayimage).This representation means that new features are extracted from the original image; this can provide us with more data than the raw information expressed as a matrix of gray values. Chapter 6 (Classification in X-ray Testing): This section covers known classi- fiers with several examples that can be easily modified in order to test different classification strategies. Additionally, the chapter covers how to estimate the accuracy of a classifier using hold-out, cross-validation and leave-one-out approaches. Chapter 7 (Simulation in X-ray Testing): This chapter reviews some basic concepts of the simulation of X-ray images, and presents simple geometric and imaging models that can be used in the simulation. Chapter 8 (Applications in X-ray Testing): This section describes relevant applicationsforX-raytestingsuchastheinspectionofcastingsandwelds,baggage screening, quality control of natural products, and inspection of cargos and elec- tronic circuits. Who Is This Book For Thisbookcoversanintroductiontocomputervisionalgorithmsthatcanbeusedin X-raytestingproblemssuchasdefectdetection,baggagescreening,3Drecognition, quality control of food products, and inspection of cargos and electronic circuits, among others. This work may not be ideal for students of computer science or electrical engineering who want to obtain a deeper knowledge of computer vision (for which purpose there are many wonderful textbooks on image processing, pattern recognition, and computer vision1). Rather, it is a good starting point for undergraduateorgraduatestudentswhowishtolearnbasiccomputervisionandits application in problems of industrial radiology.2 Thus, the aim of this book is to cover complex topics on computer vision in an easy and accessible way. 1Seeforexample[1–8]. 2Obviously,thealgorithmsoutlinedinthisbookcanbeusedinsimilarapplicationssuchasglass inspection[9]orqualitycontroloffoodproductsusingopticalimages[10]—tonamebutafew.
Description:The list of books you might like
