Ellipse fitting using finite rate of innovation principles

摘要: We address the problem of parameter estimation an ellipse from a limited number samples. develop new approach for solving fitting by showing that x and y coordinate functions are finite-rate-of-innovation (FRI) signals. Uniform samples modeled as sum weighted complex exponentials, which we propose efficient annihilating filter technique to estimate parameters The FRI framework allows estimating reliably partial or incomplete measurements even in presence noise. efficiency robustness proposed method is compared with state-of-art direct method. experimental results show estimated have lesser bias error reduced 5-10 dB relative