Robust and Efficient Ellipse Fitting Using Tangent Chord Distance

Ellipse fitting is a fundamental problem in computer vision which has been extensively studied during the past decades. However, this problem still remains unresolved due to many practical challenges such as occlusion, background clutter, noise and outlier, and so forth. In this paper, we introduce a novel geometric distance, called Tangent Chord Distance (TCD), to formulate the ellipse fitting problem. Under the least squares framework, TCD is used as the measure to quantify the fitting error, based on which a nonlinear objective function is established and minimized via the Gauss-Newton method. Compared to existing geometric distance based methods, a key merit of our approach is that, the very time-consuming iterative procedure of finding the counterparts of the given points has a simple closed-form solution in our TCD-based formulation, which can thereby significantly reduce the computational load without sacrificing the performance. Experimental results on both synthetic data and public image datasets have demonstrated the superiority of our method over other compared methods in terms of robustness and efficiency.