|
Abstract:
|
Astronomical images are highly complex arrangements of light gathered from multiple, often physically distinct but overlapping components at a variety of scales. Analyzing these images is further complicated by the presence of measurement effects like the telescope's point spread function, the detector's energy sensitivity, and the particular manner of data collection, whether as intensities measured in a pixel, or as a table of photon arrival events. It is of general interest to be able to segment these images such that each segment is focused on one component, which can then be analyzed in different ways depending on the science question that is being asked. However, while the detection of point sources has been well studied, there does not yet exist a method that can determine where the segment boundaries fall when extended sources or point sources superposed on extended sources are present. Here I describe efforts to develop a general algorithm that can be applied to images from high-energy astronomy that are based on seeded region growing concepts.
|
