A Source Classification Algorithm for Astronomical X-ray Imagery of Stellar Clusters Dr. Susan Hojnacki Chester F. Carlson Center for Imaging Science Rochester Institute of Technology Rochester, New York Monday, October 10, 2005 - 2:00 pm The Chandra X-ray Observatory (Chandra) is producing images with outstanding spatial resolution using low-noise, fast-readout CCDs. Among many other things, X-ray images and spectra help astronomers study star formation and galactic evolution. Currently, X-ray astronomers classify one X-ray source at a time by visual inspection and use of model-fitting software. This approach is useful for studying the physics of bright individual sources but is time consuming for analyzing large images of rich fields of X-ray sources, such as stellar clusters. Objective and efficient techniques from the fields of multivariate statistics, pattern recognition, and hyperspectral image processing, are needed to analyze the growing Chandra image archive. An image processing algorithm has been developed that orders the given X-ray sources based on hard versus soft X-ray emission and then groups the ordered X-ray sources into clusters based on their spectral attributes. The algorithm was applied to imaging spectroscopy of the Orion Nebula Cluster (ONC) population of more than 1000 X-ray emitting stars. As an initial test of the algorithm, images of the ONC from the Chandra archive were analyzed. The final spectral classification algorithm was applied to a sample of sources selected from among the more than 1600 X-ray sources detected in the Chandra Orion Ultradeep Project. Clustering results have been compared with known optical and infrared properties of the population of the ONC to assess the algorithm's ability to identify groups of sources that share common attributes.