基于ID-ELM算法的医药大输液可见异物检测系统研究

针对250 m L及以上医药大输液生产过程中造成药液中出现毛发、漂浮物、玻璃屑等可见异物的在线实时检测难题,开发了一种基于机器视觉的医药大输液可见异物智能检测系统。在研究医药大输液图像滤波、异物分割的基础上,设计了大输液图像特征提取算法,提出了基于影响度剪枝的改进极限学习机(ID-ELM)分类算法对可见异物分类识别,最后进行了相关实验验证算法的可行性。所设计的检测系统在医药企业用户运行并测试表明,该系统具有高检测精度、高效率、高稳定性的特性,识别准确率超过95.5%,能有效剔除次品,解决了医药大输液可见异物的在线自动检测和识别的难题,为医药生产企业分析产品质量提供了技术保障.

Development of visible foreign matter detection system for medical large infusion based on ID ELM algorithm

According to online real time detection problem of 250 mL and above medical large infusion in the production process, which caused the emergence of hair, floating debris, glass chip and other visible foreign matter, an intelligent detection system based on machine vision for visible foreign matter in medical infusion was developed. On the basis of the research for the image filtering and foreign matter segmentation of medical infusion, the feature extraction algorithm of large infusion image was designed, and pruning based on influence degree for improved extreme learning machine (ID ELM) classification algorithm was proposed for classification and identification of visible foreign matter. Finally, the feasibility of the algorithm was verified by experiments. The detection system which was tested and run by the medical enterprise users demonstrated that the system not only had high accuracy, high efficiency and high stability, but also achieved 95.5% recognition accuracy, and could effectively eliminate the defects. The problem of on line automatic detection and recognition of the visible foreign matter in medical large infusion was solved. It provided technical support for the analysis of product quality in the pharmaceutical production enterprises.