Constrained Least Squares Fitting of Ellipse

This paper presents a new efficient algorithm for fitting an ellipse to scattered points. By minimizing the algebraic distance subject to the constraint of length of semi-minor axis, the new approach incorporates the ellipticity constraint into the normalization factor. The proposed method guarantees an ellipse-specific solution even for the noised data spaning a small section of the elliptic arc. Validation with experimental results demonstrates the accuracy and the robustness of the method.