Motion of physical objects in the world is, in general, nonrigid. In r
obotics and computer vision, the motion of nonrigid objects is of grow
ing interest to researchers from a wide spectrum of disciplines. The n
onrigid objects being studied may be generally categorized into three
groups according to the degree of deformation of body parts: articulat
ed, elastic, and fluid. In articulated motion, individual rigid parts
of an object move independently of one another and the motion of the w
hole object is nonrigid in nature. Elastic motion is nonrigid motion t
hat conforms to a certain degree of continuity or smoothness. Fluid mo
tion violates even the continuity assumption and may involve topologic
al variations and turbulent deformations, This paper presents an overv
iew of existing work on articulated and elastic motion, motivated by p
roblems relating to the motion of the human body and of an animal hear
t, respectively. We study various approaches for recovering the 3D str
ucture and motion of objects through a sequence of images in a bottom-
up fashion, a strategy widely employed by various investigators. These
approaches are classified as (1) motion recovery without shape models
, and (2) model-based analysis. In the discussion of each algorithm, w
e also include a description of the complexity of feature and motion c
onstraints, which are highly related to each other. (C) 1998 Academic
Press.