基于多经验参数约束的等离子弧焊熔池边缘检测

对等离子弧焊接熔池图像边缘提取算法进行了研究.拍摄了等离子弧焊接正面熔池,发现强烈的正面电弧弧光被熔池液态金属表面反射后在图像中形成大面积的灰度饱和区,弧光区域边界的灰度梯度远大于真实熔池边缘的灰度梯度,导致常规的图像处理算法不能准确地将熔池边界提取出来.针对这一问题,提出多经验参数约束的边缘检测算法,使用前n幅已处理熔池图像边缘信息和弧光最小外接曲线等多种经验参数作为引导信息对图像边缘进行约束,剔除伪边界,检测出真实的熔池边缘信息.将所提取的边缘信息与试验测量结果进行比较,表明所提出的算法能够有效快速地检测出熔池边缘.

Plasma arc welding pool edge detection based on empirical multi-parametersconstraint

The algorithm to extractthe plasma arc welding pool edge was studied in this paper. Large area of the saturated region formed in weld poolimages because of reflected arc by weld pool surface. The gray gradient on the boundary of arc region is greater than that of the weld pool. So weld pool boundary can not accurately be extracted by the traditional image algorithms. Aiming at this problem, a multi-parameters constraint algorithm was proposed. The edge information of n frames processed weld pool images and arc minimum circumscribed curve were used as prior knowledge in the algorithm. The results show that the algorithm can effectively eliminate the pseudo boundary and weld pool edge can be detected well and quickly.