基于分布式的玻璃缺陷检测技术研究及性能优化

玻璃检测速度的的提高会在短时间会产生大量图像数据,传统分布式框架MapReduce处理速度和及时性无法满足玻璃缺陷检测的要求。课题将MapReduce分布式框架运用到海量图像处理,设计阈值分割算法完成对玻璃缺陷图像的处理。通过添加数据划分模块使计算与存储本地化,加快数据处理的及时性。实验结果表明改进的MapReduce计算框架处理速度平均提高14.1%,能够对运行速度为600m/h的玻璃带进行在线检测,并检测出玻璃带上缺陷的个数、位置和缺陷的类型。

Research and performance optimization based on distributed glass defect detection technology

With the increase of glass detection speed, some defects of MapReduce distributed computing framework are exposed, and the processing speed and timeliness cannot meet the requirements of industrial glass defect detection technology. Based on the MapReduce parallel computing framework, the paper designs a threshold segmentation method to complete the segmentation of glass defect images. By adding a streaming data processing module and a data partitioning module, the computing and storage are localized, and the timeliness of data processing is accelerated. The experiment results show that the improved MapReduce computing framework has an average processing speed increase of 14.1%. It can detect the glass ribbon running at 600m/h and detect the number, position and type of defects on the glass ribbon.