基于数学形态学BGA焊点分割和提取的改进方法

A Improvement Method On Segmentation And Extraction of BGA Joint by Mathematical Morphology

对经X射线采集到的BGA焊点缺陷图像进行预处理,传统的迭代阈值和灰度拉伸方法对灰度图像很难得到满意的结果,误分割很常见,针对BGA焊点图像的特性,提出了一种结合数学形态学的图像分割改进方法,分割后图像不仅保留了图像的重要信息,而且对于背景复杂、光照不均匀的图像也能取得良好的效果,该方法通过正确选取结构元素加快了图像分割的速度,在此基础上,采用连通域进行区域提取的方法能够准确无误的提取缺陷。实验结果表明,该方法很好实现了缺陷区域的分割和提取,且受噪声影响很小,为进一步的缺陷特征参量的处理奠定了良好的基础。

The traditional iteration threshold and gray stretching method are hard to get a satisfactory result and inaccuracy segmentation for gary image when preliminary processing to BGA joint defection image from X-ray. A improvement method combine mathematical morphology is to be advanced based on characteristic of BGA joint image. After segmentation, the image have image important information and have a good effect to the image of complicated background and asymmetric. The mathed have a faster rapid for image segmentation via right structure element. The defection are accuracy extracted from district extraction. From the experimental result, the method achieve the segmentation and extraction to defection and have a litter infection to noise. The method settle a good basement for defection processing farther.