A Discrete Particle Swarm Optimization to Estimate Parameters in Vision Tasks

Abstract

Abstract-The majority of manufacturers demand increasingly powerful vision systems for quality control. To have good outcomes, the installation requires an effort in the vision system tuning, for both hardware and software. As time and accuracy are important, actors are oriented to automate parameter's adjustment optimization at least in image processing. This paper suggests an approach based on discrete particle swarm optimization (DPSO) that automates software setting and provides optimal parameters for industrial vision applications. A novel update functions for our DPSO definition are suggested.The proposed method is applied on some real examples of quality control to validate its feasibility and efficiency, which shows that the new DPSO model furnishes promising results.

An ant colony based model to optimize parameters in industrial vision

“…Genetic algorithms [9] are widely used for this purpose. We proposed an optimization method based on particle swarm optimization [2] to find the best values of free algorithm parameters used in image processing. Further and based on this previous work, we reconsider a model that finds the best combination of operators, besides the best values of their parameters based also on a population heuristic [10].…”

“…This study is a part of an assembly of models [2,3] designed to automate the parameters setting process in image processing. The main objective is to provide a new model based on the ACO approach beside those already proposed: firstly, to compare their performances.…”