Abstract
In this paper, a new approach is proposed to extract an ordered sequence of curvilinear structures in images, capturing the largest and most influential paths first and then progressively extracting smaller paths until a prespecified size is reached. The results are demonstrated both quantitatively and qualitatively using synthetic and real-world images. The method is shown to outperform comparator methods for certain cases of noise, object class, and scale, while remaining fundamentally easier to use due to its low parameter requirement.