A Curve Finding Algorithm

• There is a large standard literature on finding curvilinear features in images.
• Following Anderson and Balasubramanian (PRD60, 1999), we intend to implement a "second derivative hysteresis threshold" algorithm.
• step 1: smooth features
• step 2: look for high curvature
• step 3: connect the dots
• have already started implementing algorithm, with improvements, with graduate student Charlie Torres as part of Kudu project, with collaborators John Baker (NASA Goddard), Manuela Campanelli (UTB), Jolien Creighton (University of Wisconsin - Milwaukee), Carlos Lousto (UTB) and Charlie Torres (UTB).