Acquisition Et Reproduction D'images Couleur: Approches Colorimétrique
Total Page:16
File Type:pdf, Size:1020Kb
Acquisition et reproduction d’images couleur : approches colorimétrique et multispectrale Jon Hardeberg To cite this version: Jon Hardeberg. Acquisition et reproduction d’images couleur : approches colorimétrique et multispec- trale. Interface homme-machine [cs.HC]. Télécom ParisTech, 1999. Français. tel-00005657 HAL Id: tel-00005657 https://pastel.archives-ouvertes.fr/tel-00005657 Submitted on 5 Apr 2004 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Thèse Presentée pour obtenir le grade de docteur de l’Ecole Nationale Supérieure des Télécommunications Specialité : Signal et Images Jon Yngve Hardeberg Acquisition et reproduction d'images couleur : approches colorimétrique et multispectrale Acquisition and reproduction of colour images: colorimetric and multispectral approaches ENST 99 E 021 Soutenue le 15 janvier 1999 devant le jury composé de Roger D. Hersch Président Jean-Marc Chassery Rapporteurs Jean-François Le Nest Hans Brettel Directeurs de thèse Francis Schmitt Kadi Bouatouch Examinateurs Robert Sève Ecole Nationale Supérieure des Télécommunications Hardeberg,JonYngve. Acquisitionandreproductionofcolourimages:colorimetricandmultispectralapproaches. Adissertationsubmittedinpartialfulfillmentofthedegreeof ”Docteurdel’EcoleNationaleSuperieur´ edesTe´ lecommunications”,´ Paris,France,1999. Keywords: Colourimaging,colorimetry,colorimetriccharacterisation,multispectralimaging, spectralcharacterisation,filterselection,spectralreconstruction. Anelectronicversionofthismanuscriptwithcolourillustrationsismadeavailableonlineat http://www-isis.enst.fr/Kiosque/theses/MANUSCRITS/Jon_HARDEBERG.pdf. Moreinformationaboutthedescribedresearchcanbefoundat http://www.tsi.enst.fr/˜hardeber/, http://hardcolor.virtualave.net/, bycontactingProf.FrancisSchmitt([email protected]), Dr.HansBrettel([email protected]), Dr.JonYngveHardeberg([email protected], [email protected]), orthroughtheaddressbelow. EcoleNationaleSuperieur´ edesTe´ lecommunications´ Departemen´ tTSI 46,rueBarrault F-75634ParisCedex13 France Acknowledgements The work described in this document has been carried out at the Signal and Image Processing De- partment at the Ecole Nationale Sup´erieure des T´el´ecommunications (ENST) in Paris, France. This endeavour would not have been completed without the support and help of many people. You all deserve thanks! Running the risk of forgetting someone important, I will mention some of you in particular. First of all, I would like to thank my advisors Francis Schmitt and Hans Brettel. Thanks for show- ing me the way into the wonderful world of colour, and for guiding and encouraging me throughout my tenure at the ENST. A big thank you goes then to all my colleagues and friends at the ENST. To the former and present Ph.D. students with whom I have shared moments of work and pleasure: Anne, Bert, Dimitri, Florence, Genevi`eve, Jean-Pierre, Jorge, Lars, Maria-Elena, Mehdi, Raouf, Selim, So- phie, Sudha, Wirawan, Yann, and many others. You have meant a great deal to me, and I sincerely hope to be able to stay in touch with you! To those of you who have worked with me on colourful projects: Bahman, Brice, Fr´ed´eric, Henri, Ingeborg, Jean-Pierre. To our relations in the industry, for showing me that my research could be able to solve some real problems out there. I thank the members of the jury, Jean-Marc Chassery, Jean-Franc¸ois Le Nest, Khadi Bouatouch, and Roger D. Hersch, for your fruitful questions, comments and suggestions, and for honoring me by your presence at my defence. Furthermore, I would like to thank the organising committees of the international conferences I have attended, for accepting the papers I submitted. Not only has this allowed me to present my work, and learn about the work of other researchers in the field of color imaging, it has given me the opportunity to meet and discuss with several inspiring individuals. I have seen that the world of colour is a small one, after all. Funding has been provided by the Norwegian Research Council (Norges Forskningsr˚ad). Its finan- cial support is of course gratefully acknowledged. I am also very grateful to Dan and RanDair at Conexant Systems, Inc. for giving me a great opportunity to continue to work in this enchanting field. Finally, I cannot imagine doing anything without you, Kristine. Thank you for giving me a reason to move to Paris, thank you for putting up with me through times of hard pressure, thank you for my two lovely children, Pauline and Samuel, and most of all, thank you for being you. Kirkland, Washington, USA November 1999 Abstract The goal of the work reported in this dissertation is to develop methods for the acquisition and re- production of high quality digital colour images. To reach this goal it is necessary to understand and control the way in which the different devices involved in the entire colour imaging chain treat colours. Therefore we addressed the problem of colorimetric characterisation of scanners and printers, provid- ing efficient and colorimetrically accurate means of conversion between a device-independent colour space such as the CIELAB space, and the device-dependent colour spaces of a scanner and a printer. First, we propose a new method for the colorimetric characterisation of colour scanners. It consists of applying a non-linear correction to the scanner RGB values followed by a 3rd order 3D polyno- mial regression function directly to CIELAB space. This method gives very good results in terms of residual colour differences. The method has been successfully applied to several colour image acqui- sition devices, including digital cameras. Together with other proposed algorithms for image quality enhancements it has allowed us to obtain very high quality digital colour images of fine art paintings. An original method for the colorimetric characterisation of a printer is then proposed. The method is based on a computational geometry approach. It uses a 3D triangulation technique to build a tetra- hedral partition of the printer colour gamut volume and it generates a surrounding structure enclos- ing the definition domain. The characterisation provides the inverse transformation from the device- independent colour space CIELAB to the device-dependent colour space CMY, taking into account both colorimetric properties of the printer, and colour gamut mapping. To further improve the colour precision and colour fidelity we have performed another study con- cerning the acquisition of multispectral images using a monochrome digital camera together with a > ¿ set of à carefully selected colour filters. Several important issues are addressed in this study. A first step is to perform a spectral characterisation of the image acquisition system to establish the spectral model. The choice of colour chart for this characterisation is found to be very important, and a new method for the design of an optimised colour chart is proposed. Several methods for an opti- mised selection of colour filters are then proposed, based on the spectral properties of the camera, the illuminant, and a set of colour patches representative for the given application. To convert the camera output signals to device-independent data, several approaches are proposed and tested. One consists of applying regression methods to convert to a colour space such as CIEXYZ or CIELAB. Another method is based on the spectral model of the acquisition system. By inverting the model, we can es- timate the spectral reflectance of each pixel of the imaged surface. Finally we present an application where the acquired multispectral images are used to predict changes in colour due to changes in the viewing illuminant. This method of illuminant simulation is found to be very accurate, and it works well on a wide range of illuminants having very different spectral properties. The proposed methods are evaluated by their theoretical properties, by simulations, and by experiments with a multispectral image acquisition system assembled using a CCD camera and a tunable filter in which the spectral transmittance can be controlled electronically. Resume´ Le but de ce travail est de d´evelopper des m´ethodes sp´ecifiques pour l’acquisition et la reproduction d’images num´eriques de tr`es haute qualit´e colorim´etrique. Pour parvenira ` ce but, il est n´ecessaire de maˆıtriser toute la chaˆıne du traitement de l’information couleur. La premi`ere partie de cetteetude ´ porte plus sp´ecifiquement sur le probl`eme de la caract´erisation colorim´etrique des scanners et des imprimantes, en nous r´ef´eranta ` un espace colorim´etrique ind´ependant : l’espace CIELAB. L’algorithme propos´e pour la caract´erisation colorim´etrique d’un scanner est le suivant : une mire de couleurs standardis´ee est d’abord num´eris´ee, puis la r´eponse RVB du scanner est compar´ee avec les valeurs colorim´etriques CIELAB de chaqueechantillon ´ de la mire. A partir de ce jeu de donn´ees, nous mod´elisons la r´eponse du scanner par une m´ethode de r´egression polynomiale d’ordre 3. Une des originalit´es de notre approche est d’optimiser