A MULTISCALE VISION MODEL ADAPTED TO THE ASTRONOMICAL IMAGES

The analysis of the sky shows many kinds of hierarchically distributed objects. We have introduced a multiscale vision model based on the wa velet transform. The discrete transform is performed by the a trous al gorithm which furnishes an isotropic vision, with a unique wavelet fun ction. The vision model is based on the notion of the significant stru ctures. We identify the pixels of the wavelet transform space (WTS) we can attribute to the objects. At each scale a region labelling is don e. An interscale connectivity graph is then established. Connected tre es are identified from the preceding graph. An object is generally ass ociated to a subtree built from this graph. The identification of WTS pixels related to a given object leads to reconstructing an image by p artial restoration algorithms. The object properties are extracted fro m the restored image. The main difficulty lies in the object reconstru ction knowing the wavelet coefficients in the volume where the object is defined. It is a classical inverse problem. We choose to solve it u sing iterative algorithms. These algorithms give correct restored imag es, as we show on different examples, without or with adding a Gaussia n noise. The influence of close objects can be partially removed.