How does the CLEAN-SC algorithm work?

CLEAN-SC (Clean based on Source Coherence) is an algorithm that is widely used for afterward-applied corrections of acoustic maps. In general, the algorithm minimizes the influence of the array-dependent point-spread function (PSF) on a previously calculated map so that various sources can be clearly distinguished from each other and especially from PSF-dependent side lobes, also known as "ghost sources". The algorithm's basis is the use of the spatial coherence between actual sources and side lobes allowing the supression of side lobes. For this, the following procedure is applied at each frequency until a stop-criterion is fulfilled:

Determine the global maximum of the uncorrected acoustic map ("dirty map") assuming that it implies the position of the most dominant source.

  1. Calculate the coherent parts of this point.
  2. Subtract these coherent parts from the "dirty map".
  3. Enter the detected maximum into an empty acoustic map ("clean map").
  4. In the end, this procedure leads to a visibly "cleaned" map which allows the identification of such sources that have been masked by main lobes of other, more dominant sources or their side lobes (see Figure 1).

CLEAN-SC minimizes the influence of the PSFs on the acoustic map. Because of its significant improvements, it is commonly used for beamforming investigations in the frequency domain. The algorithm performs less well in the case of measurements in wind tunnels or outdoor measurements at large distances from sources because of a coherence loss. Furthermore, calculation errors might occur if the resolution and/or image field are not chosen adequately. As a consequence, potential maxima cannot be identified correctly.

Further Reading

Sijtsma, P.; Snellen, M (2016). High-Resolution CLEAN-SC. Berlin Beamforming Conference. Berlin.
Sijtsma, P. (2007). CLEAN based on spatial source coherence. International Journal of Aeroacoustics, Vol. 6.

Visit the website Berlin Beamforming Conference held by GFaI e. V.