Arcfinder: An algorithm for the automatic detection of gravitational arcs

Abstract

We present an efficient algorithm designed for and capable of detecting elongated, thin features such as lines and curves in astronomical images, and its application to the automatic detection of gravitational arcs. The algorithm is sufficiently robust to detect such features even if their surface brightness is near the pixel noise in the image, yet the amount of spurious detections is low. The algorithm subdivides the image into a grid of overlapping cells which are iteratively shifted towards a local centre of brightness in their immediate neighbourhood. It then computes the ellipticity for each cell, and combines cells with correlated ellipticities into objects. These are combined to graphs in a next step, which are then further processed to determine properties of the detected objects. We demonstrate the operation and the efficiency of the algorithm applying it to HST images of galaxy clusters known to contain gravitational arcs. The algorithm completes the analysis of an image with 3000x3000 pixels in about 4 seconds on an ordinary desktop PC. We discuss further applications, the method's remaining problems and possible approaches to their solution.

12 pages, 12 figures

Cited by in corpus (6)

• First catalog of strong lens candidates in the COSMOS field
• Realistic simulations of gravitational lensing by galaxy clusters: extracting arc parameters from mock DUNE images
• Gravitational lenses and lens candidates identified from the COSMOS field
• Strong lensing statistics and the power spectrum normalisation
• A method to search for strong galaxy-galaxy lenses in optical imaging surveys
• The SDSS Discovery of a Strongly Lensed Post-Starburst Galaxy at z=0.766