Previous abstract Next abstract

Session 54 - Ground and Instrumentation Techniques and Catalogs.
Display session, Wednesday, June 12
Tripp Commons,

[54.19] Classification of Blended Images with an Artificial Neural Network

D. N. Thayer, J. G. Webster, J. A. Larsen, R. M. Humphreys (U. Minnesota), M. L. Nielsen (Boston U.)

We have developed a neural network classifier for separating blended stellar images from the "galaxy" and "star" classes of the Minnesota Automated Plate Scanner Catalog of the first epoch Palomar Observatory Sky Survey. The networks have been trained on image parameters taken from a set of visually classified stars, galaxies and blended images. The total set of images is taken from both O and E bandpasses in high and low galactic latitude fields. Once trained, the networks are tested with an independent set of images from the same fields to assess the success of the classifier at recognizing blended images. The results of this classifier will add a new classification, "blend," to the APS catalog in the near future.

Program listing for Wednesday