A multi-scale, multi-wavelength source extraction method: getsources

We present a multi-scale, multi-wavelength source extraction algorithm called getsources. Although it has been designed primarily for use in the far-infrared surveys of Galactic star-forming regions with Herschel, the method can be applied to many other astronomical images. Instead of the traditional approach of extracting sources in the observed images, the new method analyzes fine spatial decompositions of original images across a wide range of scales and across all wavebands. It cleans those single-scale images of noise and background, and constructs wavelength-independent single-scale detection images that preserve information in both spatial and wavelength dimensions. Sources are detected in the combined detection images by following the evolution of their segmentation masks across all spatial scales. Measurements of the source properties are done in the original background-subtracted images at each wavelength; the background is estimated by interpolation under the source footprints and overlapping sources are deblended in an iterative procedure. In addition to the main catalog of sources, various catalogs and images are produced that aid scientific exploitation of the extraction results. We illustrate the performance of getsources on Herschel images by extracting sources in sub-fields of the Aquila and Rosette star-forming regions. The source extraction code and validation images with a reference extraction catalog are freely available.