Motion segmentation and tracking using a seeded region growing method

We describe a user-guided system for the segmentation of an image sequence. A first segmentation which involves two images of the sequence is presented, as well as the tracking of its result during a number of frames. The segmentation algorithm is a Region Growing algorithm. The main segmentation feature is the motion of the objects presented in the image, which is combined with the information obtained by their intensity or colour - if neccessary. Two "post-processing" techniques are proposed to stabilize the object boundaries over time. A sense of layering is embedded in the whole process.

[1]  Rolf Adams,et al.  Seeded Region Growing , 1994, IEEE Trans. Pattern Anal. Mach. Intell..