Convolutional Virtual Electric Field External Force for Active Contours

Yuanquan Wang, Yunde Jia

Abstract: A novel external force called CONvolutional Virtual Electric Field (CONVEF) for active contours is proposed by taking the Virtual Electric Field (VEF) just as a convolution operation and by using modified distance metrics in the convolution kernel. The proposed CONVEF method possesses some desirable properties of VEF such as large capture range and being implemented in real-time by using fast Fourier transform. Meanwhile, the CONVEF snake provides much better segmentation than VEF snake in terms of noise suppression, C-shape concavity convergence, weak edge preserving, and neighbored objects separation. These advantages has been demonstrated and verified on synthetic and real images.