A multi-agent approach to edge detection as a distributed optimization problem

The purpose of the paper is to describe a multi-agent approach to edge detection as a distributed optimization problem. In this framework, edge detection is seen as a goal to be reached, expressed in terms of the minimization of an estimated edge detection error with respect to an ideal reference which is not explicitly known. Furthermore, an original method for distributed optimization is illustrated based on an initial partitioning of the image into zones corresponding to different characteristics. The initial partitioning can be further refined, based on the evaluation of the result obtained so far. Consistency between adjacent zones in the image is also taken into account. The implementation of this method as a multi-agent system is presented demonstrating the interest in using such systems for solving distributed optimization problems.