IMPLEMENTATION OF A PHOTOGRAMMETRIC RANGE SYSTEM 
Antonio Maria Garcia Tommaselli 
Head of the Department of Cartography 
Milton Hirokazu Shimabukuro 
Software Engineer 
Patricia A. Paiola Scalco 
Fernando M. A. Nogueira 
Undergraduate Students 
Sáo Paulo State University - Unesp 
Campus Presidente Prudente, S.P. 
e mail: ueppr(Seu.ansp. br 
BRAZIL 
Commission Ill, Working Group 2 
KEY WORDS: Reconstruction, Triangulation, Metrology, Vision, Automation, Close Range Digital Scanner, Three- 
dimensional Reconstruction. 
ABSTRACT 
This paper describes the implementation of a Photogrammetric Close Range system which aims near real time object 
reconstruction. The system is based on off-the-shelf video and a pattern projector. Stereo configuration is avoided due 
to the high computational costs of the matching procedure. The mathematical model of the object reconstruction and 
calibration is based on the parametric equation of the projected straight line combined with collinearity equations. The 
derivation of the mathematical model, methods for feature extraction and results obtained from simulated and real data 
are presented. A sequential approach to calibrate the projected bundle is discussed. Accuracy of 1/4 of the pixel size 
can be obtained with the 0-p grouping method and least squares line fitting. Experiments with simulated data have 
indicated 1mm of accuracy in height determination and O.3mm in the XY plane in an application where the separation 
between object and camera was 1000mm. 
1. INTRODUCTION Digital Photogrammetry seems to fulfill these needs and 
high precise close range surface measurements can be 
This paper is concerned with the current stage of a achieved. The traditional stereo configuration is 
Photogrammetric Range System which has been unsuitable to this specific task due to the computational 
developed at Unesp-Presidente Prudente, Brazil. A costs of the stereo-matching procedure. An active 
financial support of CNPq (National Research Council) pattern projector can be introduced and treated as an 
has been available. The motivation of the project are the active camera and only one digital imaging camera is 
needs in surface measurements for applications in used. This is similar to triangulation with structured light 
industry and medicine. Previous research projects were (Guisser et al, 1992). 
conducted using non-metric cameras to measure body 
surface shapes aiming clinical applications such as This paper presents the proposal and preliminary results 
diagnosis of postural problems associated with scoliosis, of a photogrammetric system composed of low cost 
mastectomy and other diseases. However, conventional hardware, based on a pattern projector, an electronic 
analogue or analytical Photogrammetry cannot be applied camera and image processing software. 
to solve properly these new tasks because near real time 
responses and full automation are required, as well as 
low cost systems. 2. FEATURE EXTRACTION 
Automation is of crucial importance because non 2.1 Methods 
specialists in photogrammetry have to deal with the 
system routinely. The surface measurements must be The pattern projected onto the surface must be designed 
accurate in order to derive high quality analysis of to make feasible image analysis in real time, for 
dimensions and/or deformations (Mitchell, 1995). example, a set of cross lines, squares or dots. In order to 
improve the correspondence process it is recommended 
368 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996 
“ses — (C (üho co nn 040 
Qs