Abstract
We describe an approach to motion estimation that enables the trajectory of a target undergoing substantial changes in motion between frames to be reconstructed from a single frame of data. The approach is based on the fact that time-sequentially obtained samples from a moving object form a set of noisy measurements of the trajectory. Approximation methods are used to construct a smooth estimate of the trajectory based on these samples. The methods considered here are least squares and smoothing splines. The order in which the spatial points in the field of view are sampled is shown to have a significant effect on how reliably motion is detected and estimated. The effect of object size and velocity on the motion estimate is also considered.
© 1986 Optical Society of America
Full Article | PDF ArticleMore Like This
J. B. Koskol
J. Opt. Soc. Am. A 8(6) 883-892 (1991)
Andrew B. Watson, Albert J. Ahumada, and Joyce E. Farrell
J. Opt. Soc. Am. A 3(3) 300-307 (1986)
T. S. Huang
J. Opt. Soc. Am. A 3(9) 1501-1503 (1986)