radius in 
subpixels 
01-06 
07-08 
09-10 
11-12 
13-14 
15-16 
17-18 
19-20 
21-22 
23-24 
> 24 
dynamic range 
in graylevels 
01 - 10 
01 - 10 
11-20 
11-20 
02 
02 
01 
02 
01 
01 
01 
21 - 30 
21 - 30 
03 
02 
04 
10 
07 
09 
15 
04 
09 
01 
04 
01 
31 - 10 
31 - -10 
03 
02 
17 
18 
20 
20 
12 
14 
01 
08 
02 
41 - 50 
41 - 50 
02 
15 
12 
20 
21 
09 
13 
07 
09 
02 
03 
01 
01 
51 - 60 
51 - 60 
07 
13 
14 
13 
08 
09 
04 
04 
01 
02 
01 
04 
61 - 70 
61 - 70 
01 
03 
04 
08 
04 
03 
04 
06 
01 
03 
71 - 80 
71 - 80 
01 
02 
07 
03 
02 
01 
05 
01 
02 
01 
81 - 90 
81 - 90 
01 
01 
01 
03 
91 - 100 
91 - 100 
101 - 1 10 
101 - 1 10 
111 - 120 
111 - 120 
01 
121 - 130 
121 - 130 
01 
131 - 110 
131 - 140 
> 140 
> 140 
Table 1: Histogram of form parameters of extrema in iield4: upper left entries for peaks, lower 
right entries for valleys 
5 Conclusions 
Computer load by measuring the features proposed does not seem to be a problem for our 
STARDENT 3000 which we have procured for this purpose. However, much research work 
remains to be done: Implementation of measurements for the parameters of the local dis 
tribution of the extrema, test and verification with a number of images from different SAR 
sensors, and redundancy assessment between these new measurements and standard features. 
In addition, the problem of SAR image segmentation on the basis of this larger set of features 
is anything else but trivial. 
The RADA11MAP project is sponsored by Ike German Ministry of Research and Technology, 
Bonn. 
References 
[Brisco 82] B. Brisco, R. Protz: Manual and Automatic Crop Identification with Air 
borne RADAR Imagery, Phologranimctric Engineering and Remote Sensing 
J8, Jan. 1982, page 101 
699