2004 
  
tor, 
the 
the 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
  
Unit: pixel(p) and meter(m) 
  
  
Method — vcx/p vey/p Vex/m Vcy/m Vcz/m 
LS 2.24 2.23 7.38 7.29 10.13 
Stein 1.65 1.52 5.48 5.07 4.57 
HK 1.62 1.49 5.49 4.97 4.52 
LW 921.67 140.75 2432.58 837.9 50.45 
GR 1.61 1.43 5.39 4.88 4.50 
CRS 1.41 1.33 4.47 4.34 4.39 
  
Table 4. Test results at check points for Image 02 
Table I,Table 2, Table 3 and Table 4 show that the CRS 
estimator is the most accurate, for both 10 meter resolution 
SPOT 1 image and 2.5 resolution SPOT 5 image the orientation 
precision at directional points is within one pixel and that at 
control points is with one and a half pixels. The next best 
estimators are the stein estimator, the ordinary ridge estimator 
with Horel-Kennad approach and the generalized ridge 
estimator, about one pixel accurate at directional points and two 
pixels accurate at check points. Among these three next best 
estimators the generalized ridge is slightly better than the 
ordinary ridge estimator with Horel-Kennad approach, and the 
latter is slightly better than the stein estimator. The least square 
estimator and the ordinary ridge estimator with Lawless-Wang 
approach are the worst, unstable and inaccurate. 
Therefore, it can be concluded that the ridge-stein estimator is 
the most accurate, reliable and stable approach for linear 
pushbroom imagery orientation. 
5. CONCLUSIONS 
The strong correlation among exterior orientation elements of 
linear pushbroom imagery causes the normal equation ill- 
conditioned and least squares estimation values no longer 
optimal. The new biased estimator, the combined ridge-stein 
estimator can effectively overcome the ill-conditioned problem 
and improve orientation precision. Experimental results confirm 
that the CRS estimator is superior to the least square estimator, 
the ridge estimator and the stein estimator in accuracy, 
reliability and stability. The combined ridge-stein estimator is a 
perfect approach for linear pushbroom imagery orientation. 
REFERENCES 
Gui, Q. M., Li, G. Z., 2002. Combined ridge with shrunken 
estimator and its application in geodetic adjustment. Journal of 
Geodesy and Geodynamics, 22(1), pp. 16-21. 
Guo, H.T., Zhang, B. M., Gui Q.M., 2002. Application of 
generalized ridge estimation in computing the exterior 
orientation elements of satellitic linear array scanner imagery. 
In: The International Archives of the Photogrammetry, Remote 
Sensing and Spatial Information Sciences, Xi'an, China, Vol. 
XXXIV, Part 2, pp. 153-156. 
Gupta, R., Hartley, R., 1997. Linear pushbroom cameras. /EEE 
Trans. PAMI, 19(9), pp.963-975. 
Huang W. B., 1992. Modern Adjustment Theory and lts 
Application, Beijing Publishing House, Beijing, pp.120 -132. 
731 
Zhang, F. R., Wang X. Q., 1989. Ridge and stein estimation for 
adjustment parameters. Journal of Wuhan Technical University 
of Surveying and Mapping, 14(3), pp. 46-57.