Ellipse Fitting Based on a Hybrid l1l2-Norm Algorithm.

This paper proposes a new efficient direct method noted the Hybrid l1l2-Norm Model for fitting ellipse to resist the Gaussian and Laplacian disturb simultaneously which will be solved by split Bregman iteration and shrink operator. The experimental results reveal that the proposed method not only works well with Gaussian Noise data but Laplacian Noise data and its mixed version. Several experimental examples are reported to demonstrate the robustness of the proposed approach.