位置信息约束的SMT料盘X射线图像检测方法

发布时间：2018-10-05 10:18

【摘要】：SMT封装物料盘的X射线透视图像发生局部粘连,导致分割准确性降低影响物料的检测与计数统计。提出了基于位置信息约束的分割检测方法,利用器件位置信息对分割区域进行约束提高分割与检测的准确率。首先依据元器件呈螺旋状排列规则,拟合物料盘中心点并计算最内环排列中起点器件位置信息;然后基于中心点与各元器件的位置约束模型,进行法向位置信息与先验位置信息双重约束,对目标器件所在区域进行限定;最后划分分割界限,完成对粘连目标的分割和对元器件的检测与统计。实验结果表明:该方法能够提高物料盘X射线透视图像粘连区域分割与检测的准确率,在有效像元素9 216 pixel,细节解析度110 1p/cm成像环境之下,对不同规格物料盘进行实测,检测误差率控制在0.15%之内。
[Abstract]:The local adhesion of X-ray fluoroscopy image of SMT packaging material disk results in the reduction of segmentation accuracy and affects the detection and counting statistics of materials. A method of segmentation and detection based on position information constraint is proposed, and the accuracy of segmentation and detection is improved by using the location information of the device to constrain the segmented region. Firstly, according to the spiral arrangement rule of components, fitting the center point of material disk and calculating the position information of the starting point in the innermost ring arrangement, then based on the position constraint model between the center point and each component, The normal position information and the prior position information are constrained to limit the region where the target device is located. Finally, the segmentation boundary is divided to complete the segmentation of the adhesion target and the detection and statistics of the components. The experimental results show that this method can improve the accuracy of segmentation and detection of adhesion area in X-ray fluoroscopic images of material disk. Under the imaging environment of 9 216 pixel, detail resolution of effective image elements, the material plates of different specifications are measured. The detection error rate is controlled within 0.15%.
【作者单位】：天津市光电检测技术与系统重点实验室;天津工业大学电子与信息工程学院;
【基金】：国家自然科学基金(61405144) 天津市科技支撑计划重点项目(14ZCZDGX00033) 天津市科技特派员项目(15JCTPJC56300)
【分类号】：TP391.41

【参考文献】