Filtering of Signal Dependent Noise Applied to MIPS Data

Linear filtering and usual nonlinear median filtering are not effective for signal-dependent noise removal. We apply here an approximate decoupling of signal and noise by means of a nonlinear transform. The transform is followed by a linear filter and the corresponding inverse transform. This procedure allows us to mitigate the signal-dependent noise in the images obtained by the Multiband Imaging Photometer for Spitzer (MIPS), 70μm imaging band.