This paper proposes a Kalman-filter-based algorithm for fitting multiple ellipses. Specifically, the curve in Cartesian coordinates is first transformed into an extended curve in polar coordinates to obtain ordered data. Then, the Kalman filter is used for clustering to select points belonging to the same candidate ellipse. Finally, the parameters of each ellipse are solved using the least squares method to achieve ellipse fitting. Overall, the proposed algorithm effectively fits multiple ellipses while avoiding a large number of ineffective samples and maintaining a low computational load. Moreover, the numerical simulation results demonstrate that the proposed method has good robustness against outliers.