The Number of Looks and InfoPACK

Many of the InfoPACK routines, for example the utility stats.py (or from the GUI, select Tools/Image Statistics), estimate the number of looks from data using the definition of ENL above. Since this estimate normally contains both RCS and speckle fluctuations this estimate only corresponds to the ENL when there are no RCS variations. We describe the result as the "looks measure" to distinguish it from the true ENL.

Occasionally, InfoPACK routines accept the number of looks as an input. In this case the defined instrument value, eg. 3 for full resolution ERS PRI, should be used.

Even if you do know the number of looks in your data, you may still get better results by allowing the InfoPACK functions to estimate it from the image. The InfoPACK functions use the looks parameter to build a model of the surface cross-section that underlies the speckled image. When the observed intensity is divided by the reconstruction the result resembles pure speckle so that the looks measure is equivalent to the ENL. It is important to get a good model, and the functions are tuned to work with their own estimates. The estimates are usually slightly different from the `real' value. They are close when the image consists mainly of fields, and get poorer when estimated from a town scene.