320 
The second test was taken from a test field in which target templates were distributed in depth. Figure 3 shows the distribution of the 
templates within the second test field. The method for template recognition is different with the first method, because the ability of 
pattern recognition using different methods was decided to test. In this test an edge enhancement mask was applied on the image and 
templates were detected by scanning a window across the image. Scanning was used instead the collinearity method in this instance 
to investigate a less constrained case. Short camera focal lengths combined with a close distance of the camera to the test field and a 
poor knowledge of the camera calibration factors for the distance can lead to positional uncertainty, and ambient lighting may lead to 
a saturation background on the templates. The size of square side was about 6 pixels on image, which is approaching the level where 
the size of templates is less than 5 pixels and numbers can not be reliably identified. So it is necessary to set up the camera at a 
distance such that the size of square side is not less than 10 pixels on the image. At this distance the background has less effect on 
the templates. The numbers are recognisable even on images which are approaching saturation by the background. For example, 
numbers on Figure 2 are more recognisable than numbers on Figure 3. 
The third test has been carried out using a digitised aerial photograph. In this test the templates were superimposed on the aerial 
photograph of an urban area using a raster image processing software package (see Figure 4). The whole of the image was scanned 
without any initial pre-processing or filtering. Despite the fact the photograph was taken of an urban area and the entire image 
included objects similar to the target templates, only the templates were recognised and detected. This test verifies that the templates 
have their own unique identity and hence are recognisable amongst other objects. 
4. CONCLUSION 
Target templates have been designed for digital close range photogrammetry, but they can be used in other disciplines of 
photogrammetry. The results were obtained from some tests in close range photogrammetry verified that the templates can be 
identified under a variety of conditions. Blurred and saturated images, caused by lighting and background conditions, are not a new 
problems in photogrammetry. However, the template design has a strong structure and relationship which can be recognised despite 
poor images. In the worst case an edge enhancement mask or sharpness mask can be applied to the image in order to improve the 
results. The parameters of camera calibration and camera position should be initially introduced if detection of the targets is to be in 
real time. 
An algorithm has been developed for the recognition and detection of the templates. The algorithm is able to define the position of 
control points and the number refer to control points. Some problems have been encountered in the processing. These problems are 
related to the condition of lighting and background saturation. Another problem is related to the size of templates on the image. To 
achieve good results it is suggested that the size of side of the square on the image is not less than 10 pixels. : 
In the aerial photogrammetry case, templates were superimposed on a digital aerial image. The image was taken of an urban area 
with the high density residential development. The algorithm can successfully recognise templates and the control points with their 
numbers among other objects. This test proves that if a good contrast is obtainable between the control points and their numbers with 
the background, the templates can be recognised and detected. 
Analysis of the results which have been taken from the above mentioned tests indicates that the templates would be recognisable in 
all types of photography. The next stage of the project is to utilise the target detection and identification to implement automatic 
stereo matching in digital close range photogrammetry. 
5. REFERENCES 
Forstner, W. and Gulch E., 1987. A fast operator for detection and precise location of distinct points, corners and centres of circular 
features. ISPRS Intercommission workshop, Interlaken, Switzerland, pp 281-305. 
Forstner, W., 1986.A feature based correspondence algorithm for image matching. Int. Arch. of Photogrammetry and Remote 
Sensing, Vol. 26-3/3, pp 150-166. 
Gruen, A., 1988. Towards real-time photogrammetry. Photogrammetria, 42, pp 209-244. 
Gruen, A., 1992. Tracking moving objects with digital photogrammetric systems. Photogrammetric Record, 14(80), pp 171-182. 
Gruen, A., 1994. Digital close-range photogrammetry - progress through automation. Proceedings, ISPRS Commission V 
Symposium on Close Range Techniques and Machine Vision. Melbourne, Australia, pp 122-135. 
Helava, U. V., 1988. On system concepts for digital automation. Photogrammetria, 43, pp 57-71. 
Homainejad, A. S., 1992. The problems of real time photogrammetry. Final thesis of Master Surveying Science, The University of 
New South Wales, Australia. 
Lemmens, M., 1988. A survey on stereo matching techniques. Proceedings, Commission 5, 16th Congress of the International 
Society for Photogrammetry and Remote Sensing, Kyoto, Japan, pp 11-23. 
Moravec, H.P., 1977. Towards automatic visual obstacle avoidance. Proceedings, 5th Int. Joint Conf. Artificial Intelligence, 
Cambridge, MA, USA, p 584. 
Nasrabadi, N. M., 1992. A stereo vision technique using curve-segments and relaxation matching. IEEE Transaction on Pattern 
Analysis and Machine Intelligence, Vol. 14, No. 5, pp 566-572. 
Sim, D. G., Ham, Y. K. and Park, R. H., 1994. On-line recognition of cursive Korean characters using DP matching and fuzzy 
concept. Pattern Recognition, Vol. 27, No. 12, pp 1605-1620. 
van den Heuvel, F. A., Kroon, R. J. G. A. and Le Poole, R. S., 1992. Digital close-range photogrammetry using artificial targets. 
Proceedings, Commission 5, 17th Congress of the International Society for Photogrammetry and Remote Sensing, Washington 
DC, U.S.A., pp 222-229. 
Vollmerhaus, D. and Bildanalyse, K., 1987. A fast algorithm for local matching of patterns in images. Proceedings, Intercommission 
Conference on Fast Processing of Photogrammetric Data Interlaken, Switzerland, pp 273-280. 
Wiley, A. G., and Wong, K. W., 1990. Metric aspects of zoom vision. Proceedings, SPIE Vol. 1395 Close Range Photogrammetry 
Meets Machine Vision, Zurich, Switzerland, pp 112-118. 
IAPRS, Vol. 30, Part 5W1, ISPRS Intercommission Workshop "From Pixels to Sequences", Zurich, March 22-24 1995 
  
era rate ra Tu rar pa EE