Automatic Control Point Measurement

Nowadays the tenor of reports on applications of digital photogrammetric procedures consists of a unison message: processing modules of automatic image orientation are available in digital photogrammetric systems and with them the productivity is much higher than with all existing conventional photogrammetric plotters. In fact, interior and relative orientation procedures as well as automated aerial triangulation have matured. Unfortunately this positive balance is not generally valid for automatic control point measurement which is required in automating exterior orientation. In this paper we review developments on automatic measurement of ground control and present an investigation on (semi-)automatic extraction of signalized points. The philosophy behind this investigation follows a pragmatic idea: for signalization of a block a simple signal structure is used which promises a high success rate in detection and location of control points. Experimentally a block of 36 photographs was processed with totally 1714 image points of imaged targets. The measurement rate of over 99% of successfully localized signals is remarkably high. Furthermore, the accuracy of this automatic measurements of 2.8 μm is 15% better than the 3.2 μm accuracy obtained for manual measurements. 1. AUTOMATION OF EXTERIOR ORIENTATION Automatic exterior orientation is dominated by research which so far diverges in two completely different directions. On one hand there is GPS and INS technology which permits direct measurement of position and attitude parameters. In principle, a multisensor camera system with integrated GPS/INS sensors produces exterior orientation parameters without any ground control which is also called geo-referencing. Consequently photogrammetric control points which primarily serve for datum transform from a local sensor system to a global geodetic coordinate system would become obsolete. That this assessment does not bear closer examination is outlined by Ackermann (1997). He concludes that the analysis of accuracy performance discloses that especially INS does not yet meet the high photogrammetric demands. Moreover, control and elimination of systematic errors and mainly the weak reliability of GPS geo-referencing make the use of some control points mandatory. On the other hand there is the traditional indirect determination of exterior orientation. Control information must be provided externally and research tries to solve the measurement of control points or structures in digital images. Performance expectation on automatic measurement defined by human operator's efficiency is high. An operator measures control points fairly quick, reliable and accurate. There is a certain dependency between achievable precision and type of control information (signalized points, selected natural points and man-made structures, e.g. buildings or street crossings) whereas highest precision is usually obtained with signalized points. An overall measurement process consists of detection, identification and precise location of imaged control points. Unfortunately, the automation of these three steps of control point measurement is more difficult than photogrammetric image measurement tasks like interior orientation or point transfer. This is the major reason why the development of algorithms for reliable location of ground control points has not shown much progress during the last decade. 2. AUTOMATION OF GROUND CONTROL POINT MEASUREMENT The general problem of measuring ground control is to establish correspondence between a given model of a control point and its representation extracted from an image. If control information is imaged in two or more photographs, multiple image matching provides restrictions and additional information for model to image correspondence. For exterior orientation a broad spectrum of 'Photogrammetric Week '97' D. Fritsch & D. Hobbie, Eds., Wichmann Verlag, Heidelberg, 1997.