Heikkinen, Jussi
CIRCULAR IMAGE BLOCK MEASUREMENTS
Jussi Heikkinen
Helsinki University of Technology
Institute of Photogrammetry and Remote Sensing
Finland
Jussi.Heikkinen@ hut.fi
KEY WORDS: Close-range photogrammetry, Video recording, Estimation, Image sequence
ABSTRACT
At the moment, more and more rapid 3D documentation and less expensive measurement methods are required in
photogrammetric reconstruction tasks. One considerable alternative will be methods utilizing video recording
implementation. In this paper we will show a method based on continuous video recording sequences and freenet
type estimation. The system is designed to be semiautomatic and to be used by non-photogrammetrists without
knowledge of photogrammetric planning and geometrical aspects of imaging.
In this paper, we will concentrate on theoretical background of the method and especially on parameterization of the
estimation problem. Results of simulated tests of image block formation will be presented. The main goals of tests are
to verify the stable conditions for circular image block formation and solve the extent and goodness of scale
measurements to be required in image block formation.
1 INTRODUCTION
The demand of 3D scene models and to get them in a fast and relatively easy way has increased lately. The consequence
is that more automatic methods in image measurement systems are required as also that the imaging system has to be
well controlled. By using retro or more preferably coded targets, the efficiency of the measurement systems can be
improved drastically. Although the time spent for preparation; targeting and planning the imaging configuration; is
considerable. Often the case is that we have numerous objects of the same kind and we can plan the imaging
configuration only once. Then we only have to target the object and refine the measurements respect to the datum.
In a case we have a unique modelling task, we have to create a co-ordinate system, plan the camera positions and
possibly do targeting before we get to the measurement part of the system. Planning the camera configuration might
take time, as we have to design the control point distribution and camera positions respect to the object very carefully in
order to achieve the required accuracy. Cause all these parts conflict each other, we have to approach this problem
iteratively. The design of the measurement system can be divided into ZOD, FOD, SOD, and TOD levels of planning
according to Fraser (Fraser, 1989). This type of procedure requires experience and is hardly a job of non-professionals.
To meet the first requirement of fast and relatively easy ways of modelling, we have to develop methods that are
reasonably automatic and can regulate the imaging geometry in order to fulfil the requirements of the modelling
accuracy. The system has to be easy to operate by non-professionals and measurement conditions have to be controlled
so that the imaging geometry will be sufficient for the modelling purpose. The system should take care of the imaging
geometry and the modelling accuracy even when the operator does not understand the concept of accuracy in
photogrammetric 3D modelling.
The method presented here will produce a reasonably good geometry for 3D measurements from multiple images. The
idea is to construct an image block like in a conventional close-range photogrammetric method. Instead of defining the
best positions of camera stations for the measurements, the camera positions will be constrained to lie on the path of the
circle. It is also required that such a circular image block will be closed i.e. first and last image have to overlap. The
imaging geometry may not be optimal for the task, but it certainly will give a sound basis for estimating the accuracy of
the resulting model.
358 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.