Improving Gamma-ray Source Searches with Fast Fourier Transforms and Image Processing

The High Altitude Water Cherenkov (HAWC) observatory surveys the very-high-energy sky in the energy range from 300 GeV to greater than 100 TeV. Its wide field of view makes it particularly suitable for studying large, extended regions of emission that often contain multiple point and diffuse sources of gamma rays. Existing blind search methods to detect sources in HAWC data use a computationally expensive, iterative source fitting algorithm that can take several days to scan a few degree region in the sky. In this work, we adopt a new approach to speed up the identification of sources using an image processing pipeline. Using image processing filters, and a blob finder algorithm, based on the Determinant of a Gaussian, this pipeline seeds sources accurately up to 300 times faster than the current HAWC source search pipeline. The pipeline's output are then passed onto a global multi-threaded likelihood fitter for accurate source localization. We present the performance of the improved pipeline and discuss prospects for future applications on other astrophysical datasets.