Abstract—An automatic system for stamp segmentation and
further verification is needed especially for environments like
insurance companies where a huge volume of documents is
processed daily. However, detection of a general stamp is
not a trivial task as it can have different shapes and colors
and, moreover, it can be imprinted with a variable quality
and rotation. Previous methods were restricted to detection of
stamps of particular shapes or colors. The method presented
in the paper includes segmentation of the image by color
clustering and subsequent classification of candidate solutions
by geometrical and color-related features. The approach allows
for differentiation of stamps from other color objects in the
document such as logos or texts. For the purpose of evaluation,
a data set of 400 document images has been collected, annotated
and made public. With the proposed method, recall of 83%
and precision of 84% have been achieved.
Keywords-stamp detection; image segmentation; computational forensics; color clustering