A Modification of the Level Set Speed Function to Bridge Gaps in Data

Abstract. Level set methods have become very popular means for image segmentation in recent years. But due to the data-driven nature of this methods it is difficult to segment objects that appear unconnected within the data. We propose a modification of the level set speed function to add a "bridging force" that allows the level set to leap over gaps in the data and segment an object despite artifacts or partial occlusions. We propose two methods to define such a force, one model-based and one image-based. Both versions have been applied to a series of test images, as well as medical data and photographic images to show their adequacy for image segmentation.

• Type: Conference
• Year: 2006
• Where: DAGM
• Authors: Karsten Rink, Klaus D. Tönnies
• Keywords: DAGM 2006, Image Processing, Image Segmentation, Level Set, Level Set Methods