i 
i 
output of information on the planimetric geo- 
metry and for the numerical description of a ter- 
rain by elevation points. 
The chairman closed the session by suggesting 
the establishment of a test field for the com- 
parison of different D.T.M. 
Thursday, July 15, 1976, from 14.00 to 15.30 
Session 5 of Commission Ill 
Topic: Use of Reseau Photography 
Chairman: Dr. H. Bauer (F.R.G.) 
Presented Papers 
Author: Mr. R. Roberts (Australia) 
Title: Test of Bundle Block Adjustment 
for Survey Co-ordination PP 111/49 
Mr. Roberts reviewed results obtained in Australia 
using a bundle block adjustment for third order 
surveys to establish control for property surveys 
utilizing 1:12700 scale photography taken with 
a camera equipped with a reseau. Tests had 
shown that sheet metal painted with a circular 
target of "blaze orange" fluorescent paint giving 
a negative size of 40 microns was the practical 
minimum size. It had been decided that all pass- 
points would be targeted and all survey control 
points would appear on 9 photographs. The over- 
lap used had been 6096x6096. A bundle type 
block adjustment program had been used. 
The results show that it is possible to obtain a 
vertical accuracy of 2—3 um and a horizontal 
accuracy of 4—5 um. 
Encouraged by the results from these tests full 
scale trials were proceeding over an area of 
900 km? at Kapunda using reseau photography 
taken from 7,500 meters above ground level. 
The presentation was followed by a lively dis- 
cussion on reseau techniques. 
Author: Mr. D. Brown (U.S.A.) 
Title: The State of Development and of 
Application of Numerical Photo- 
grammetry Methods 
Mr. Brown recommended the use of auto-regres- 
Sive methods for the analysis of statoscope data. 
With the help of inertial systems carried by jeeps 
and helicopters it is possible to obtain an ac- 
curacy of 1:220000 (0.65 m) in control point ex- 
tension. These systems could compete with 
photogrammetric methods in the future. Their 
use results in considerable savings in the num- 
ber of control points. Mr. Brown suggested that 
image coordinates be weighted radially and tan- 
gentially. Control point coordinates should no 
longer be introduced as errorless. He also re- 
commended a wider use of reseau cameras. 
Brown had developed a reseau platen of his own 
that should be more accurate than the others 
available. (49 crosses are produced with small 
projectors. With suitable measuring arrange- 
ments additional measurements present no prob- 
lem. When talking about self-calibration, Mr. 
Brown mentioned that further investigations are 
necessary. For this purpose several good test 
fields are needed. He observed that more cor- 
rection terms are required for S.W.A. photographs 
than for W.A. ones. In particular with S.W.A. pho- 
tographs a reseau can be most useful. 
In the course of the intensive discussion that 
followed, the audience gave its support to many 
of Mr. Brown's theses. 
Monday, July 19, 1976, from 15.45 to 17.15 
Session 6 of Commission Ill 
Topic: Image Processing 
Chairman: Dr. G. Konecny (F.R.G.) 
Invited Paper 
Authors: Dr. F. Holdermann (F.R.G.) 
Dr. M. Bohner (F.R.G.) 
Dr. B. Bargel (F.R.G.) 
Dr. H. Kazmierczak (F.R.G.) 
Title: Review of Image Processing /P ////05 
Presented by Dr. Bargel 
A review of image processing was given in which 
various methods for radiometric and geometric 
corrections, image enhancement, feature extrac- 
tion, and classification for feature analysis were 
discussed. 
At the close of this presentation, Dr. F. Leberl 
(U.S.A.) asked whether the processing of visual 
images should not have been given a greater 
emphasis than automatic feature extraction. 
Dr. Bargel answered that although he himself 
has concentrated his efforts on feature extraction, 
the paper here given is not one-sided. 
Presented Papers 
Author: Dr. P. Anuta (U.S.A.) 
Title: Digital Preprocessing and Classifi- 
cation of Multispectral Earth Obser- 
vation Data PP 111/09 
In Dr. Anuta’s paper a computer software system 
concept called LARSYS for preprocessing and 
analysis of multivariate image data from remote 
sensing measurements was described. The sys- 
tem includes statistical analysis, data classifica- 
tion, image registration, geometric corrections, 
and ratioing principle components. 
Topographic data which had been registered with 
four band LANDSAT multispectral scanner data 
was being studied to develop layered classifier 
and other methods of utilizing elevation, slope 
and aspect to improve the quality of multispectral 
image analysis. 
In the course of the discussion that followed, 
Dr. A. McNair (U.S.A.) remarked that, also at his 
institute, a strong correlation had been observed 
between channels 4 and 5 and channels 6 and 7 
of the LANDSAT imagery. In New York, the tem- 
poral display had been carried out with four 
channels using the G.E. Image-100 equipment. 
Dr. E. Mikhail (U.S.A.), who read Dr. Anuta’s paper, 
observed that due to high correlation only two 
channels had been used. 
Dr. A. Fontanel (France) asked whether the in- 
formation on slope had improved the classifica- 
tion results. Dr. Mikhail answered that the results 
had been somewhat improved. 
Author: Dr. G. Kritikos (F.R.G.) . 
Title: Image Processing and Multispectral 
Classification at the DFVLR 
Dr. Kritikos gave a paper on image processing 
and multispectral classification at the DFVLR. 
103