Application of moment and Fourier descriptors to the accurate estimation of elliptical-shape parameters

Abstract Accurate estimation of the parameters of an elliptical shape is required in various machine-vision and computer-vision problems. In our previous work, we have addressed this problem by proposing to optimize a weighted minimum-squares-error (MSE) function. As a continuation of this work, we have studied other techniques for elliptical-parameter estimation, ones applying elliptical-Fourier descriptors, moments of area, and moments of perimeter. A study, to be reported here, was carried out to evaluate the comparative performance of the above-mentioned three techniques and the one based on the weighted MSE function. The limitation and degree of accuracy of each technique was determined. It was found that different elliptical-parameter-estimation techniques must be applied depending on acceptable computational cost, number of parameters to be estimated, the required degree of accuracy, and the specific conditions under which the estimation must be performed.