Multiscale Edge Detection in the Corona

Coronal Mass Ejections ( CMEs ) are challenging objects to detect using automated techniques, due to their high velocity and diffuse, irregular morphology. A necessary step to automating the detection process is to first remove the subjectivity introduced by the observer used in the current, standard, CME detection and tracking method. Here we describe and demonstrate a multiscale edge detection technique that addresses this step and could serve as one part of an automated CME detection system. This method provides a way to objectively define a CME front with associated error estimates. These fronts can then be used to extract CME morphology and kinematics. We apply this technique to a CME observed on 18 April 2000 by the Large Angle Solar COronagraph experiment ( LASCO ) C2/C3 and a CME observed on 21 April 2002 by LASCO C2/C3 and the Transition Region and Coronal Explorer ( TRACE ). For the two examples in this work, the heights determined by the standard manual method are larger than those determined with the multiscale method by ≈ 10% using LASCO data and ≈ 20% using TRACE data.