Imaging of planets is very difficult, due to the glare from their nearby, much brighter suns. Static and slowly-evolving aberrations are the limiting factors, even after application of adaptive optics. The residual speckle pattern is highly symmetrical due to diffraction from the telescope's aperture. We suggest to break this symmetry and thus to locate planets hidden beneath it. An eccentric pupil mask is rotated to modulate the residual light pattern not removed by other means. This modulation is then exploited to reveal the planet's constant signal. In well-corrected ground-based observations we can reach planets six stellar magnitudes fainter than their sun, and only 2-3 times the diffraction limit from it. At ten times the diffraction limit, we detect planets 16 magnitudes fainter. The stellar background drops by five magnitudes.
The trouble with planetsPlanets in other stellar systems (exoplanets) are difficult to discover, and even more difficult to image. This is because there is a very large intensity difference between them and their parent stars, which can be anything from 10 -4 in the infrared down to 10 -10 and fainter in the visible regime, depending on the physical properties of the system [1,2]. Light from the star is scattered off-axis by the atmosphere, the aperture of the telescope, the secondary mirror holders (spiders) and other optical elements in the path to the imaging camera. Techniques such as adaptive optics (AO) and coronagraphy aim to remove this effect by concentrating light back on the optical axis and removing it [3]. All future ground-based high-contrast imaging systems will employ coronagraphs and AO [4,5].It has been shown through simulations and experiments that the limiting factors in highcontrast imaging from ground and space are static and quasi-static speckles [6][7][8]. While residual atmospheric aberrations after AO correction are random and will average out over time, these persistent speckles will stand out against the AO-corrected stellar halo and masquerade as faint sources even after long integrations. This is why future efforts to directly image planets will also entail post-processing schemes based on the concept of point spread function (PSF) subtraction. Proposed approaches utilize PSF estimates provided by on-sky rotation [9], as well as spectral [10,11] and polarization-based [12] discrimination between the light coming from the parent star and the companion. The problem with some of these techniques is that the PSF-subtracted images still contain static speckles (at a lower brightness level than in the direct images) due to errors in PSF estimation. These errors arise due to the inherent sensitivity of PSF subtraction to changes in "seeing", mechanical flexure or the introduction of extra imaging channels. Recently a detection algorithm has been proposed which relies only on multiple exposures and statistical properties of AO-corrected intensity [13].