基于序贯相似性与光源自动调节的芯片表面缺陷检测算法.docVIP

基于序贯相似性与光源自动调节的芯片表面缺陷检测算法 　　摘 要： 为了解决当前缺陷检查算法在保护罩反光环境下易导致采集芯片连接器图像光照不均，使其缺陷检测精度较低，无法精确定位出浅插与破损的不足，设计了基于序贯相似性检测匹配与光源自动调节的芯片连接器表面缺陷检测识别算法。首先，基于序贯相似性检测匹配，定位出连接器位置，提取出连接器不良常发区域；然后设计了光源自动调节机制，视觉软件通过光源数字控制器SDK与网线触发光源关闭或调暗，消除保护罩反光对图像质量的影响；最后采用线检测，统计缺陷区域二值图特征，完成缺陷检查。实验数据显示，与当前缺陷检查技术相比，在面对保护罩反光严重环境下的芯片连接器缺陷检查时，该算法具有更高的检查精度。 　　关键词： 缺陷检测； 图像处理； 光源自动调节； 序贯相似性检测； 线检测 　　中图分类号： TN911.73?34； TP391 文献标识码： A 文章编号： 1004?373X（2017）05?0058?05 　　Abstract： Since the current defect detection algorithm used in the protection cover reflective environment is easy to result in the nonuniform illumination of the acquired image of the chip connector， make its defect detection precision low， and is unable to locate the shallow insert and damage， an algorithm for chip connector′s surface defect detection based on the sequential similarity detecting marching and light source automatic adjustment was designed. On the basis of the sequential similarity detection matching， the position of the connector is located to extract the often occurring nonuniform illumination area of the connector. The light source automatic adjustment mechanism was designed. The visual software triggers the light source′s turnoff or dimming through the light source digital controller SDK and cable to eliminate the effect of light reflected by the protection cover on image quality. The line detection is used to count the features of the binary image of the defect area to detect the defect. The experimental data shows that， in comparison with the current defect detecting technology， the algorithm proposed in this paper has higher detecting precision while detecting the defect of the chip connector in the protection cover reflective environment. 　　Keywords： defect detection； image processing； light source automatic adjustment； sequential similarity detection； line detection 　　0 引 言 　　?S着“中国制造2025”概念的深化，将智能化、自动化技术应用于制造业，从而提高制造水平、生产效率，降低人力成本，成为非常迫切的需求[1?2]。芯片连接器是电子制造业的核心部件，产量高、工艺复杂，对后续表面缺陷检查有很大的挑战，而表面缺陷检查直接关系到产品质量，因此将智能化技术应