一种多尺度分形的弧焊熔池图像分割方法
A multi-scale fractal image segmentation method for arc welding pool
-
摘要: 弧焊熔池图像的各个区域与最终的焊缝成形有着密切的关系.分析了弧焊熔池图像中的浮渣与熔池的分型维数和多尺度分型维数的特征,阐述了图像的分形维数和多尺度分形维数特征的概念,介绍了单一分形维数与基于多尺度分形维数的图像分割方法,并把该方法应用到了弧焊熔池图像的分割中.结果表明,基于分形维数的多尺度图像分割方法可以有效地对弧焊熔池图像中熔池尾部和浮渣同时进行分割,并优于用算子和单一的分形维数算法的图像分割方法.基于分形维数的多尺度图像分割算法在弧焊熔池图像的分割上将会成为十分有效的方法.Abstract: An individual section of the arc welding pool image has a close relationship with the ultimate weld seam formation. An analysis on the fractal dimension and multi-scale fractal feature of the slag and the weld pool part is performed,and the definitions of graphical fractal dimension and multi-scale fractal feature are also elaborated. An image segmentation method using both single fractal dimension algorithm and multi-scale fractal feature-based algorithm is introduced,which has been applied in the processing of arc welding images. Experiments show that the tail of the weld pool and the slag are effectively divided in the way mentioned above,which is superior to the segmentation method based on Sobel operator and single fractal dimension algorithm.