ABSTRACT In the context of the SPHERE planet “nder project, we further develop and characterize a recently proposedmethod for the ecient direct detection of exoplanets f rom the ground using spectral and angular dierentialimaging. The method, called ANDROMEDA, combines images appropriately into pseudo-dataŽ, then uses allof them in a Maximum-Likelihood framework to estimate the position and ”ux of potential planets orbitingthe observed star. The methods performance is assesse d on realistic simulations of images performed by theSPHERE consortium, and it is applied to experimental data taken by the VLT/NAOS-CONICA instrument.Keywords: exoplanets, detection, inverse pro blems, coronagraphy, angular imaging, dieren tial imaging, spec-tral imaging, high contrast, maximum likelihood, adaptive optics. 1. INTRODUCTION The direct detection of exoplanets from the ground is a very promising “eld of astronomy today. A goal is thecharacterization of the physical composition of the exoplanets by a spectral analysis of their emitted and/orre”ected light. This observation from the ground is a t echnological challenge. I ndeed, in order to be able toobserve a sucient number of targets, it is required to cope with an intensity ratio (also called contrast) betweenthe star and its planet that may be as high as 10