Problemy automatycznego modelowania i teksturowania obiektów opisujących skrajnię linii kolejowych

The goal of the paper is to present results of research that has been conducted for several years in the Department of Geoinformatics, Photogrammetry and Environmental Remote Sensing, AGH University of Science and Technology in the field of automatic modelling and texturing of objects that describe railway line clearance gauges. In the paper, the authors have presented bases of 3D modelling and texturing of objects, with particular consideration of railway clearance gauge objects. An attempt has been made at assessing possibilities of automatic and semi-automatic reconstruction of objects located in the direct vicinity of railway tracks on the basis of a point cloud from a mobile scanning obtained, together with digital images from four cameras, for the selected test railway line section on Cracow-Warsaw route. Railway objects shape construction gauge of railway lines, therefore virtual geometric model of those objects might be used for checking if railway rolling stock elements of specified dimensions can safely fit the gauge outline. A couple of programs have been tested in the framework of research (PhotoModeler Scanner, RiScan Pro, 3DReshaper) as regards their usability for modelling and texturing. In research works a previously prepared triangle model has been utilized, as well as a set of 160 images made with the use of four cameras. Additional data included camera and distortion parameters, and elements of external orientation of obtained images. The paper shows exemplary results and indicates problems originating in the course of creating models by means of point cloud vectorization. The authors' experiences show that this is a far more difficult process than stereoscopic photogrammetric model vectorization. Only an experienced operator, with a good spatial imagination, is capable of producing a correct model. That is why in majority of solutions, semi-automatic methods are applied. Those methods consist in the operator's determining of a type (or, possibly, an approximated shape) of the object to be detected, as well as its rough location, and it is the task of the algorithm to fit the object into a set area. The conducted research has demonstrated that there is no thoroughly satisfactory method (program) for automatic modelling and texturing of railway line clearance gauge. As of today, numerous projects have still to be performed in either a semi-automatic, or a manual way.

W artykule podjęty jest problem automatycznego modelowania i teksturowania obiektów kształtujących skrajnię kolejową. Źródłem modelowania jest mobilny skaning laserowy, a teksturowania – zdjęcia wykonane podczas skaningu. Artykuł składa się z dwóch części: przeglądu literatury i opisu eksperymentu badawczego. Eksperyment badawczy potwierdził większość problemów artykułowanych w literaturze, wniósł także szereg nowych. Przestrzeń obiektów przytorowych była bardzo trudnym materiałem do wymodelowania i pokrycia teksturą fotograficzną, co uwypukliło problemy. W konkluzji wskazano, że główną przeszkodą w automatyzacji procesu teksturowania jest słaba jakość modeli generowanych automatycznie.