quality of the final estimates, but in this particular 
application the contrast method offers some specific 
advantages. This is because the contrast method inherently 
uses a set of velocities over which the contrast is 
maximized, and this results in a simple algorithm as we will 
see below. 
Figure 1. shows the method for estimating the 
velocity as a function of time. The image data is divided 
into blocks in the along-track direction, i.e. the time 
axis. Each such block is then processed with a set of 
different velocities, and the contrast is calculated for 
each block and each velocity. The result is a set of 
contrast values, where each box in figure 1. corresponds to 
one contrast value. For each step in time the maximum 
contrast value is then found, where the maximization is done 
over the set of velocities. A possible outcome of this 
operation could be the set of maximum contrast values shown 
in figure 1., where each darkened box corresponds to the 
maximum value for the contrast curve in question. The exact 
estimate is then found by interpolation in each contrast 
curve. The final set of velocities as a function of time is 
then found by applying a short lowpass filter where the 
actual values of the contrast maxima are used as weights. 
This is because a high value of the contrast peak is 
considered more significant than a lower value. 
VELOCITY 
  
TIME 
Fig.1. Estimation of velocities from set of 
maximum contrast values. 
396