摘要
设计了一套基于机器视觉技术的全自动灯检系统,并提出了一种弱小异物目标识别算法,以实现大输液中可见异物的在线实时自动检测。为了有效地解决在线高速实时检测这个关键性问题,将制约系统实时性的弱小目标识别过程分离出来,交由高速DSP芯片进行专门处理。通过理论分析和试验测试证明,基于TMS320C6416 DSP的图像处理平台能够很好地满足大输液异物检测系统对处理速度和精度的要求。该系统能够有效地检测输液中粒径大于50μm的异物,其识别准确率和速度均高于熟练的灯检工。
To detect automatically the rare objects on line in infusion procedure, an algorithm for identifying small rare objects of particulate matter is proposed, and a detection system based on machine vision technology is designed. In order to resolve this critical problem of effective detection in high speed and real time, the identification of small objects is processed independently by high speed DSP chip. Through theoretical analysis and experimental test, the design of image processing platform based on TMS320C6416 DSP is able to meet the requirement of speed and accuracy for rare object detection. The system can detect the rare particulate matter that diameter is greater than 50 μm, and the identification accuracy is similar to that from skilled operators for regular lighting detection.
出处
《自动化仪表》
CAS
2008年第8期22-25,共4页
Process Automation Instrumentation
基金
山东省自然科学基金资助项目(编号:Z2006F05)。
关键词
异物检测
机器视觉
DSP
输液灯检
图像处理
形态学滤波
Rare object detection
Machine vision
DSP
Infusion inspection
Image processing
Morphological filtering
