Level Set Based Shape Model for Automatic Linear Feature Extraction from Satellite Imagery

Although image linear feature is important in many applications, it is difficult to extract precisely, because the background is complex and feature is not always significant. In this paper, the authors propose a level set based automatic linear feature extraction method using prior shapes. The authors introduce a variable shape model which together with the level set function for segmentation and extract linear feature dynamically indicates the region with which the prior shape should be compared. The model is capable of extract linear objects from an image based on the image intensity as well as the prior shape. In addition, the proposed model permits translation, scaling and rotation of the prior shape. Moreover, a fast way is also established for the minimization of the function. At last, the experiments validate the model, and a comparison with canny detector showed the advantage.