Spectral Subtraction: A New Approach to Remove Low‐ and High‐Order Speckle Noise

Abstract

We present a novel ''spectral subtraction algorithm'' (SSA) technique to remove speckle noise. It consists of a loworder and a high-order SSA and is based on a three-dimensional image spectroscopy in which the three-dimensional data cube is available and thus the speckle noise introduced by the wave-front error can be efficiently subtracted. For the low-order SSA, speckles up to the second or third order can be totally subtracted, leaving the residual speckles dominated only by the third or fourth order, respectively, and imaging contrast is increased consequently; for the high-order SSA, speckles up to the fourth or fifth order can be subtracted, leaving the residual speckles dominated only by the fifth or sixth order, respectively, and the performance is further improved. This is the first demonstration that such high-order speckles could be subtracted. Since the SSAs are conducted over a wide spectral band, a whitelight image can be re-assembled from the three-dimensional data cube. The white-light image would increase the single-to-noise ratio and reduce the exposure time, which are crucial for the search of faint companion objects. Combined with a coronagraph, the SSA can provide an extra contrast gain for the coronagraph imaging, relax the requirement for the wave-front quality (no adaptive optics correction is required for a space-borne imaging system), and significantly increase the performance of exoplanet imaging and biomarker spectroscopy.

Speckle Noise Subtraction and Suppression With Adaptive Optics Coronagraphic Imaging

“…This well-known speckle n 1 is discussed by many authors (Perrin et al 2003;Bloemhof 2004Bloemhof , 2007Ren & Wang 2006).…”

“…The properties of speckle noise have been discussed by many authors (Racine et al 1999;Sivaramakrishnan et al 2002;Perrin et al 2003;Bloemhof 2004Bloemhof , 2007Ren & Wang 2006). In general, the pupil wave-front error or phase Φ of an electromagnetic wave can be expressed as a convergent Taylor series e iΦ = 1 + iΦ − Φ 2 /2 − iΦ 3 /6 + Φ 4 /24 + .…”

“…A full expansion of the speckles for the aberrated point-spread function (PSF) intensity was discussed by Perrin et al (2003). For a wave-front or phase error Φ and a normalized pupil function A, the focal plane PSF intensity distribution up to the fourth-order speckle was derived in our previous publication (Ren & Wang 2006), which is suitable for an unapodized pupil. A coronagraph can use an apodized pupil to attenuate the diffraction light from an aperture, so that high contrast is available in a discovery area.…”

Estimated performance of a symmetric nulling coronagraph for exoplanet imaging

“…The residual speckle noise induced by the wave-front error can be further subtracted by using the Spectral Subtraction Algorithm (SSA) ( The SSA can provide an extra contrast gain of 10 -4 ~ 10 -5 at the 0.9 Strehl ratio (Ren & Wang 2006). Fig.…”

“…Ren & Wang 2006) that is a further development of differential imaging(Marois et al 2000). For space-based exoplanet imaging for the detection of life, there are a number of biomarkers that have strong spectral features, which could be used for differential imaging.…”