ACQUIRING INITIAL VALUE OF MULTI-VIEW METRICAL DATA INTEGRATION 
Zheng Li 1 Zhang Jianqing 1 Yuejun Luo 2 
1.Institute of Remote Sensing Information Engineering, Wuhan University, 129 Luoyu Road,Wuhan,China,430079 
2.Institute of Resource and Environment, Wuhan University, 129 Luoyu Road,Wuhan,China,430079 
KEYWORDS: Initial value, CCD camera, Projector, Reference Frame, Calibration, Rotating Grid Table 
ABSTRACT: 
Acquiring the initial value of all profile of the object is essential for the integration of metrical data. It is described a new smart 
approach to acquire the initial value of all profile of the object in this paper, which is only to compute the calibration parameters of 
CCD camera and the projector in every angle, but not to do the complicated coordinate conversion. In the origin position the CCD 
camera and the projector are calibrated by using the relation of 2D-DLT and collinearity equation orderly. And the CCD camera is 
calibrated in every visual angle by the rotating grid table. While the positions of the camera and the digital projector are both never 
removed in the whole process, the calibration parameters of the projector are computed successfully through the transformation 
matrix. The experimental results indicate that the proposed method is helpful to the multi-view metrical data integration in the 3D 
measurement. 
1. INTRODUCTION 
The CCD camera and the projector are to be calibrated before 
3-D measurement, and then it is not adjusted for the interior and 
exterior orientation elements of the CCD camera and the 
projector again. For the limitation of the view range of 
measurement, which does not meet the need of the whole object 
measurement, there are two solutions to resolve the problem. 
One makes use of the rotating table, the other dose not. There 
are two methods by using the rotating table. One is obtained the 
exact position by the precise servo mechanism, where vision 
sensor is fixed on and its measurement pose is adjusted 
scheduled orientation by detecting and programming of 
multi-sensors. The other is offered by the rotating table and 
revised by the software, where the object is rooted the rotating 
table and rectified the relative position with the vision sensor by 
turning the rotating table. It is presented the binocular 
stereovision system based on CCD camera and projector, in 
which the exterior orientation elements of CCD camera and 
projector in every angle can be calculated by making use of 
rotating grid table. Then it makes it possible to multi-view data 
integration, but avoids coordinate rotation and initial match. 
This paper is structured as follows: in section 2, reference frame 
is presented. Then section 3 explains the new smart approach to 
acquire the initial value of point cloud in every side of the 
model. Section 4 presents some experimental results and section 
5 discusses the conclusion. 
2. REFERENCE FRAME 
2.1 System Configuration 
In this paper, the 3-D measurement system is composed of five 
components: a charge coupled device (CCD) camera, a digital 
projector, rotating grid table, an object to be measured, and a 
personal computer with an image grabbing and processing 
board, as shown in Fig.l. The rotating grid table can be 
controlled by the personal computer to rotate arbitrary angle, 
which are flexible, precise and fully software programmable. 
CCD Camera 
Fig.l reference frame in the measurement system 
2.2 Reference Frame 
The whole 3D measurement process is not separated from the 
reference frame. Taken Fig.l as an example, there are generally 
five reference frames. 
1) Reference frame in original position of the rotating grid 
table 
Reference frame in original position of the rotating grid table 
OYY7 
¡ s reference frame in zero position, where the center 
of the rotating grid table is the origin of the reference frame. 
Then the axis of X and Y constitute the plane which is in 
the rotating grid table. The direction of Y ax j s j s vertical to 
the plane each other. 
2) Reference frame in arbitrary position of the rotating grid 
table 
Reference frame in arbitrary position of the rotating grid table 
is reference frame of the rotating grid table which is turned to an 
angle. The new reference frame is obtained from circling the 
7 (1XY7 
axis of ^ in ^ and rotating an angle. In the mergence 
of traditional metrical data, the metrical data, which is got by 
rotating certain angle on the rotating grid table, is to be 
transformed to reference frame in original position. And every 
model data is corresponding to every rotating angle of rotating 
grid table. Therefore coordinate transformation is needed to 
741