金属零件直接成形熔积层轮廓识别与缺陷分析

Contour Recognition and Defect Analysis for Metallic Parts Deposition Shaping

为提高金属零件熔积直接成形的形状尺寸精度和性能质量的可控性,需要实时监测熔积层宽度和识别缺陷及其形状特征。基于CCD数字图像,开发了动态二次生长算法提取真实的熔积层边缘。为解决提取的边缘不连续问题,开发了过滤边缘突变元素的衔接算法。最后,采用基于最大熵的图像分割方法提取缺陷区域,并求取了缺陷区域的位置和形状特征,为缺陷修复提供依据。实验结果表明,该方法能够准确对熔积层宽度进行可靠监测,并对缺陷区域进行准确定位和形状描述。

To assure the controllability of the dimensional accuracy and the performance quality for metallic parts deposition shaping,it is necessary to monitor the deposition width and identify the defect position.A dynamic twice growth algorithm is developed based on CCD digital image to extract the true weld edge.In order to avoid the incontinuity of the weld edge,a linking algorithm is developed to filter those unstable elements.Finally,by using image segmentation method based on maximum entropy,defect areas are extracted and those location and shape feature are calculated,both of which will offer aids for repairmen of the defects.Experiments concerned indicate algorithms proposed can exactly monitor the width of deposition layer,identify the defect area and calculate the parameters of the shape and location of those defects.