Level Set Based Simultaneous Background Image Modeling and Foreground Segmentation
Sukho Lee, Moon Gi Kang · 2007
We propose a level set based energy functional that deals with the problem of background mod-eling and object segmentation in a coupled way such that the two processes mutually constrain each other, and from which a level set based Euler-Lagrange equation can be derived. Due to the mutual constraining, the temporal change of the scene is reflected in the construction of the reference background image quickly, while a clean background image can be obtained with a relatively small number of frames. Furthermore, due to the direct implementation of the Euler-Lagrange equation, the algorithm works in real-time. PROPOSED MODEL We search for the background image B and the segmented image sequence I(t) as the solution that minimizes the following energy functional: E(B, φ) =