Affiliations: [a] University of sciences and technology of Oran Mohamed BOUDIAF, Oran, Algeria | [b] University of Mascara, Mascara, Algeria
Correspondence:
[*]
Corresponding author: Abdelkader Fekir, University of sciences and technology of Oran Mohamed BOUDIAF, USTO-MB, BP 1505 EL M'Naouer, Oran 31000, Algeria. Tel./Fax: +213 45 813 998; E-mail:aekfekir@univ-usto.dz/aekfekir@gmail.com
Abstract: This paper proposes a new multi agent system for boundary detection and
object tracking in image sequence. It exploits advantages of three elements:
multi agent solution, active contour model and multi resolution treatment.
The proposed system is organized as multi level structure where each level
is composed of many cooperative agents. In edge detection step, snake agents
in higher level of image pyramid cooperate to get object boundaries in low
resolution image. Then, it transmits its result to subsystem agents of
inferior level and these ones cooperate to refine object edge in next image,
and so on. In the second step, high level agents migrate to the next pyramid
for tracking edges in the next image of sequence, in parallel with detection
step of inferior subsystem in previous image, using the same cooperation and
scheduling strategy. The efficiency of the system is shown through some
experimental results.
Keywords: Multi agent system, active contour, image sequence segmentation, edge detection, tracking
