Abstract
This paper presents a new methodology for the design
of image processing algorithms. The presented approach
is based on the conception and design of "image
exploring agents". Image exploring agents iteratively
run a sense-compute-act loop. While all loop processing
is performed within the image on a local base, the
trace of the agent, which results from the repetition
of this loop, is a global image property. Hence, a
given global recognition task is represented by a
corresponding task of finding a suitable, local-based
computation. The exploitation of genetic programming
for finding suitable computational parts of the agent
is considered. A case study is included, in which a
crack detector is designed and improved by means of an
image exploring agent. The functionality of this agent
is studied, and it is used for designing an image
processing algorithm for crack detection. Also, further
improvements of the presented agent concept are given
in this paper. c 1999 Elsevier Science B.V. All rights
reserved.
Users
Please
log in to take part in the discussion (add own reviews or comments).