/ORKING 
ia 
ith estimation of 
t technology of 
nt technology in 
stimation of tree 
> and calculation 
during conveyor 
g a 3D model of 
uisition and the 
nation includes a 
ame grabber and 
) measurements. 
d discussed. The 
package images 
1d approach uses 
obtain rectified 
. This technique 
ipsoidal shapes 
loped algorithms 
hey are based on 
d computational 
o. 
ted non-contact 
of a log moving 
ne determination 
ing its movement 
ting a set of the 
o generate a 3D 
timation. Special 
providing metric 
proposed as an 
| requirements to 
rements 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
e support of manual mode for 3D measurements 
e capability of measurement of area of every cut tree in 
package 
e capability of measurement of area of polygon including all 
cut trees in package 
e capability of calculating the number of trees in package 
e capability of modification into real-time measurement 
system 
Below the developed methods of 3D measurements are 
described along with outlines of the hardware used for image 
acquisition and the software realized the developed algorithms. 
2. SYSTEM OUTLINE 
2.1 Hardware configuration 
The proposed volume measurement techniques were developed 
using laboratory photogrammetric system operating with model 
wood logs and wood packages of the scale factor 1:10 (the real 
size of objects to be measured being in range of 2.5-3.0 m.). 
The laboratory system is based on PC as central processing unit 
and non-metric CCD video camera as image acquisition device. 
It includes: 
e  Pentium4-1.4/512MB PC 
e 2 high resolution PULNiX CCD cameras equipped with 12- 
mm focal length lenses 
e 2 channel PCI frame grabber 
e Laser stripe line projector 
The PC is equipped by frame grabber providing simulations two 
channel image acquisition and on-board stripe light detection 
function. The view of laboratory system is shown in the Fig. 2. 
  
Figure 2. The view of laboratory system 
The system is designed for working space of 300x300x200 mm 
dimensions. 
2.2 System calibration 
As long as laboratory system uses non-metric cameras for 
measuring purposes the preliminary cameras calibration is 
performed. The original calibration procedure based on images 
acquisition of planar test field is fully automated as a result of 
applying coded target for reference point marking (Knyaz V., 
1998; Knyaz V .,2002 ). 
Calibration includes two stages: a) interior camera parameter 
determination and b) system external orientation. As a 
consequence of first calibration stage the parameters of interior 
orientation (principal point x, v,, scales in x and y directions 
m, m,, and affinity factor a, the radial symmetric Kj, KK; 
distortion and decentering P, P; distortion) are estimated. 
Precision results (standard deviation) of interior orientation 
estimation are given in table 1: 
  
spatial reference points coordinates 0,,7 0.011 mm 
angle exterior orientation parameters Oye = 0.038° 
Opp = 0.034 mm 
  
  
  
  
  
residuals of collinearity conditions 
  
Table 1. Precision of interior orientation 
The exterior orientation procedure (second stage) is performed 
using the same planar test field for estimating unknown relative 
orientation parameters: (X;, Y, Z;) — location and (04,0,K;) 
and angle position of the left camera and (X,, Y,. Z,) — location 
and (05,05,&;) and angle position of the right camera relatively 
external coordinate system origin defined by test field. The 
residual of collinearity conditions for external orientation 
procedure is 0.04 mm. 
The results of calibration demonstrate the reasonable accuracy 
for task of 3D measurements. 
To support the proposed technology original software for 
WindowsXP is developed. It provides the complete technology 
of 3D measurement: 
* image acquisition 
e System automated calibration 
* image processing for feature extraction 
e calculating the required characteristics and report 
generation 
3. PACKAGE AREA MEASUREMENT 
3.1 Measurement techniques 
Two approaches for automated front area of a package 
measuring are developed. The first approach is aimed on 
recognising all ellipses (log cuts) in two images and establishing 
correspondence between the found circles. Then area of spatial 
circle can be determined. This approach has problems for 
convergent image acquisition when the circle looks like ellipse 
(Fig. 3.) and the problem of ellipse recognition is more 
complicated then circle extraction.