image for this case also. In this case the output. image will 
be geometrically distorted in the along-track direction by 
the correction process, and the final image is obtained by 
reinterpolating in the distorted image. In case of severe 
cross-track irregularities, the target histories will be 
distributed over several range bins, and this range 
migration must. be. removed in order to ’obtaiñ a fully 
compressed image. 
3. TEST OF CORRECTION METHOD. 
In order to test the method described above, a test 
has been performed with real SAR data. Data from the SEASAT 
SAR system were used in order to obtain a controlled test. 
The data set was warped according to a selected velocity 
history by interpolating in the original data, thereby 
artificially generating data from an irregularly moving 
platform. By applying the correction method to this data set 
and comparing the results with the deliberately introduced 
velocity variations, the performance of the method was 
assessed. 
The time-bandwidth product for the reference filter 
with SEASAT SAR is approximately 4000, and in order to make 
the data set more similar to data obtained from an airborne 
SAR system, the data were lowpass filtered in along-track 
direction. The resulting reference filter time-bandwidth 
product is approximately 200. 
RELATIVE VELOCITY 
T 
1.042 + 
EXPECTED VELOCITY e 
ESTIMATES FROM 
^ \ 
1.028 4 / X 
7 % 
\ WARPED DATA 
1.014 — 
1.000 — 
  
0.986 
VELOCITY ESTIMATES 
FROM UNWARPED DATA 
0.972 — 
  
Figure 2. Focusing results from unwarped data and 
velocity history introduced. 
398