Discrimination of point-like objects in astronomical images using surface curvature

Laboratoire d'Astronomie Spatiale du CNRS, BP. 8, F-13376 Marseille Cedex 12, France

Send offprint request to: A. Llebaria

Received: 14 April 1997
Accepted: 4 June 1997

Abstract

We present a method for the discrimination of point-like objects in astronomical images using surface curvature. The principal curvatures k₁ and k₂ are calculated using a local quadric approximation to the log of the intensity of the original image. The analysis of the diagram allows to easily separate various shapes and to establish a criterion for discrimination. A mask is then generated and the invalid pixels are replaced by a local, pyramidal interpolation using a non-linear, multiresolution method. We present two applications, the discrimination and removal of stars from the dust tail of comet P/Halley observed while it was crossing the Milky Way, and the removal of cosmic ray impacts from images of the solar corona obtained with a space-coronagraph.