An improved GVF snake model and its application to linear feature extraction from SAR images

In this paper, an improved GVF snake model that allows controllable snakes is proposed. Two kinds of extern constraint forces are exploited in the model. The first one can pin specified points on the snake and determine the basic shape of a snake. The second one avoids generating ears during curve evolution. It ensures that the curves are smooth and won´t grow in a wrong direction. The improved snakes are employed to close gaps in linear feature extraction since they can fixes the connection points during the deformation and provide smooth linking curves rather than straight lines. The experimental results of ridge (and ravine) extraction and road extraction from real SAR images increase the correctness and quality of extracted results.