用遗传算法的阈值选取法检测铝合金MIG焊熔池边缘

针对铝合金MIG焊熔池边缘提取问题,建立视觉传感和采集的实验系统,提出基于最大类间方差阀值的遗传算法来提取铝合金MIG焊熔池边缘.介绍系统与算法的结构,同时与基于Sobel算子、Canny算子的熔池边缘提取结果进行对比.结果表明,遗传算法阀值图像分割在熔池图像边缘提取是可行的,通过该算法提取的铝合金MIG焊接熔池边缘清晰,可以有效地克服熔池图像信号中的噪声及阴极雾化区的影响,该方法提供了一种新的焊接熔池边缘实时检测与控制的方法.

Detection of welding pool edge of aluminum alloy MIG welding with genetic algorithmic threshold selection

Aimed at the problem of extraction of welding pool edge of aluminum alloy MIG welding,an experimental system of vision sensing and acquisition was set up,and an inter-category square deviation based threshold value selection method was proposed for extraction of welding pool edge of aluminum alloy MIG welding.The configuration of system and algorithm were introduced and,meanwhile,its result of welding pool edge extraction was compared with that of Sobel-operator and Canny-operator based method.It was shown by this comparison that threshold-value image dividing with genetic algorithm was feasible in the edge extraction of welding pool image.The edge of welding pool of aluminum alloy MIG welding extracted with this algorithm was clear,and the influence of noise in the signal of welding pool image as well as the influence of the cathode mist could effectively be eliminated by using this algorithm,providing a new real-time detection and control method for welding pool edge.