Research of Immune-Ant Colony Algorithm for the Detection of Weld Pool Image

Abstract This paper has researched a new immune-ant colony algorithm for weld pool detection.Firstly, the tungsten inert gas welding experiment system has been setup and welding images can be captured in real time. The tungsten inert gas welding experiment system is firstly built up, which can be used for images real time capturing. Based on the experiment system, the immune-ant colony algorithm has been designed.In the end, several testing experiments have been performed, which dictated that the performer of the researched algorithm is better. In testing experiments, after about 35 times iterative computations, a global optimum value for image segmentation is determined,which is 40% faster convergence compared to standalone ant colony algorithms. And wonderful weld pool segmentation images can be received. The edge of pool and weld seam can be easily extracted, which is helpful for weld automation.