Full text: Photogrammetric and remote sensing systems for data processing and analysis

L= 
1t 
rs 
LS 
1S 
[— 
l7 
1e 
4 
re 
Le 
t 
|= 
3.3 Azimuth Compression Module with Storage of Result 
An alternative concept for the SAR azimuth compression in time domain 
has been proposed by G. Kluge (ref. 7). As can be seen from equation 
[5], a fixed range compressed sample contributes to a large number of 
resulting samples. The idea is to keep the whole result area (range 
length times azimuth matched filter length) in memory, to perform for an 
input sample all multiplications and to accumulate the results to the 
corresponding positions. 
In order to achieve the necessary speed to following concept of parallel 
processing is foreseen (figure 4): 
- there are L*l parallel processing elements, each with sufficient memo- 
ry to hold the results of one range column and all necessary coeffi- 
cients; 
- a range compressed input sample is fed simultaneously by means of a 
common data bus to all processing elements; 
- each element is dedicated to a fixed position in azimuth while pro- 
cessing input from the same range column; 
- in each processing element a new range compressed sample is multiplied 
with the corresponding coefficients as indicated in equation [5 and 
accumulated to the memory positions given by the RCM offset. Using 
third order interpolations this implies normally four adjacent posi- 
tions, however the range dependence of the RCM causes occasional steps 
which can be forced to be periodic, but the period being different 
from one azimuth position to the next; 
- in order to avoid movement of data, the processing elements change 
their role cyclically whenever a new range column starts. One spare 
processing element is used for the conversion to magnitudes for a com- 
pleted column and to save the data. 
input of range compressed samples simultaneously to all 
processing elements 
die 
qo 
  
  
  
  
  
i 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
processor processor processor processor 
n n+l n+e n+3 
à P 
data | |coef. data| |coef. data||coef. data||coef. 
mem. | |mem. mem. | |mem. mem. | |mem. mem. | |mem. 
  
  
  
  
4 
  
  
! 
i» 
  
Po 
result azimuth compressed sample output from the current 
processor, e 
spare 
.g. number 
ntl, 
Figure 4: Azimuth compression module version B with 
output buffering. The data memory of each 
  
processing element has the capacity of one range column. 
The processors change with gach new range column cycli- 
cally their role with respect to azimuth position matched 
filter. 
443 
 
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.