An Objective Function for 3-D Motion Estimation from Optical Flow with Lower Error Variance Than Maximum Likelihood Estimator

Norio Tagawa, Takashi Toriu, Toshio Endoh

Abstract: This paper describes a new objective function for estimating 3-D motion from an optical flow. We construct a generalized Rayleigh-quotient form objective function, which provides an unbiased estimator and whose use drastically reduces the computational cost. This function includes an arbitrary weighting function. We show by numerical evaluations that with a suitable weighting function it provides an estimator whose variance is smaller than that of the maximum likelihood estimator.< >