We present a general framework for determining probability distributio
ns over the space of possible image feature groupings. The framework c
an be used to find several of the most probable partitions of image fe
atures into groupings, rather than just returning a single partition o
f the features as do most feature grouping techniques. In addition to
the groupings themselves, the probability of each partition is compute
d, providing information on the relative probability of multiple parti
tions that few grouping techniques offer. In determining the probabili
ty distribution of groupings, no parameters are estimated, thus elimin
ating problems that occur with small data sets and outliers such as th
e compounding of errors that can occur when parameters are estimated a
nd the estimated parameters are used in the next grouping step. We hav
e instantiated our framework for the two special cases of grouping lin
e segments into straight lines and for grouping bilateral symmetries w
ith parallel axes, where bilateral symmetries are formed by pairs of e
dges. Results are presented for these cases on several real images. (C
) 1996 Academic Press, Inc.