3-5-8 
By analyzing Table 2, the following conclusions can be 
drawn: 
1) In Guangxi project, without any ground control point, 
the practical coordinate accuracy of combined bundle 
block adjustment with GPS-determined 3D coordinates 
of the camera positions are ± 11,072m for planimetry and 
± 2.572m for height in the ground. Analyzing the 
residuals of the check points, we see that the remarkable 
systematic errors exist in GPS camera stations. In spite of 
the fact that the GPS camera coordinates possess certain 
systematic errors, yet the adjustment method has reached 
relatively high accuracy, which can meet the specification 
of topographic mapping at 1:50 000 scale. Consequently, 
it is feasible to update the topographic maps and 
topographic mapping at small or medium scale by GPS- 
supported aerotriangulation without ground control point. 
2) If 4 ground full control points around the comers and 
a few elevation control points in two sides of the site are 
given, the coordinate accuracy of the GPS-supported 
bundle adjustment reaches 1.2a 0 ~3.6cr 0 for planimetry 
and 2.0a o ~4.4a o for height in the theory and 1.6a 0 ~4.3a 0 
for planimetry and 1.7ct 0 ~3.1ct 0 for height in ground. The 
adjusted results are very closed to that of the 
conventional bundle adjustment with 3 additional 
parameters. Consisting of 4 XYZ ground control points 
near the comers of block area the model correcting the 
drift errors can be introduced into the combined bundle 
adjustment. The processing of combined bundle 
adjustment can automatically separate the systematic 
errors of the GPS camera positions and make them retain 
solely the accident errors. The residual accident errors 
can be distributed by the least squares method. Therefore, 
the accuracy of the bundle adjustment showed significant 
improvement. In this case, moreover, comparing with the 
conventional bundle adjustment, the work amount of field 
survey is reduced 88% and the production cost decreased 
75% in the current standard of the cost accounting for 
Chinese photogrammetry. 
3) The measuring accuracy of image point observations 
waves between ± \ \2fjm and±24.8//m, which depends 
upon aerophotographic quality. The GPS-supported 
aerotriangulation accuracy will be improved if a Q of 
image coordinates further reduces. 
5. SUMMARY 
The above analysis and discussion have shown that the 
combined bundle adjustment with GPS-determined 
camera stations leads to an extension of the conventional 
bundle adjustment model without posing any major 
problems. The experiments show that introducing highly 
precise kinematic camera positions plus 4 XYZ ground 
control points into the combined bundle adjustment 
results in sufficient accuracy. Due to varying systematic 
errors the combined bundle adjustment computed without 
any ground control point will most probably be limited to 
low accuracy requirements. However, it is shown from 
the experiments that kinematic GPS relative camera 
positioning for aerotriangulation is a highly operational, 
robust and economic method which can thoroughly 
change aerial photogrammetry within a short time. 
ACKNOWLEDGMENTS 
This work would not have been possible without the help 
of many people and the support of many units. The 
authors would like to express their hearty gratitude to 
Professor Jiyu Liu and other research team members. We 
also want to explicitly thank China National Bureau of 
Surveying & Mapping, Wuhan Technical University of 
Surveying & Mapping, National Geomatics Center of 
China, China Siwei Surveying & Mapping Technology 
Corporation, China Aviation Remote Sensing Services 
Corporation, Tianjing Academy of Surveying & Mapping, 
Heilongjiang Bureau of Surveying & Mapping, Shanxi 
Bureau of Surveying & Mapping, Hainan Bureau of 
Surveying & Mapping and Sichuan Academy of 
Surveying & Mapping for their energetic support. In 
addition, this project is supported by Development 
Foundation of Surveying & Mapping, China National 863 
Hi-tech Projects (No. 863-308-13-04(2)) and Natural 
Science Foundation of China (No. 49631050). 
REFERENCES 
[1] Ackermann, F., 1991. GPS for Photogrammetry, Proceedings 
of Tutorial on "Mathematical Aspects of Data Analysis ”, 
ISPRS, ICWG III/VI, pp. 17~70. 
[2] Forstner, W., 1988. The Reliability of Block Triangulation, 
Photogrammetric Engineering & Remote Sensing, 51(6): 
pp.l 137-1149. 
[3] Friess, P., 1991. Aerotriangulation with GPS-Methods, 
Experience, Expectation, Proceedings of 43 rd 
Photogrammetric Week, Stuttgart University. 
[4] Kubik, K., 1982. An Error Theory for Danish Method, 
Proceedings of the Symposium of Commission III of the 
ISPRS, Helsinki. 
[5] Li, D.R., and Yuan, X.X., 1995. GPS-supported Bundle 
Block Adjustment-An Empirical Results from Test Field 
Taiyuan (in Chinese), Acta Geodaetica et Cartographica 
Sinica, 24(2): pp.l~8. 
[6] Lucas, J.R., 1987. Aerotriangulation without Ground Control, 
Photogrammetric Engineering & Remote Sensing, 53(3): 
pp.311-314. 
[7] Wang, Z.Z., 1990. Principle of Photogrammetry (with 
Remote Sensing), Publication House of Surveying and 
Mapping, Beijing. 
[8] Yuan, X.X., 1994. GPS-supported Combined Bundle Block 
Adjustment Based on Differential Carrier Phase 
Observations (in Chinese), M.S Dissertation, WTUSM. 
[9] Yuan, X.X., 1995. The Processing of GPS System Offset On 
GPS-supported Bundle Block Adjustment (in Chinese), 
Journal of WTUSM, 20(3): pp.l 98-201. 
[10] Yuan, X.X., 1996. Correction of Earth Curvature in GPS- 
supported Bundle Block Adjustment (in Chinese), Journal of 
WTUSM, 21(3). 
[11] Yuan, X.X., and Li, D.R., 1997. Some Investigations for 
GPS-supported Aerial Triangulation (in Chinese), Acta 
Geodaetica et Cartographica Sinica, 26(1).