摘要
提出了一种利用机器视觉技术对溶液中μm级异物微粒进行检测并统计粒径信息的新方法。首先,建立溶液离心旋转急停后其中异物粒子的运动轨迹数学模型;然后,提取溶液视觉图像序列中每个可能目标的有效特征,通过特征匹配得到帧间若干可能目标的运动轨迹,根据异物目标与背景噪声等伪目标轨迹的差异进行甄别检测;最后,对视觉系统进行标定,确定检测到的异物目标粒径大小及各区间微粒数量。实验表明,该技术检测精度能够到达10μm,且具有较高的检测准确率。
A novel particle detection technology for foreign micro based on machine vision is proposed,which can detect and measure micro-scale targets in solution.Firstly,mathematical model of the particles′ centrifugal motion trajectory is established.Then,extract the effective features of potential targets from image sequences;through matching these features of adjacent frames,the potential motion trajectories of targets can be attained.In accordance with the trajectory differences between particles and pseudo-targets such as background and noise,the existence of the particles can be determined.The final calibration of the vision system is to identify the size and the number of the particles detected.The experiments show that detection precision using the proposed technology can be reached 10 microns with high detection accuracy.
出处
《光电子.激光》
EI
CAS
CSCD
北大核心
2010年第9期1346-1349,1364,共5页
Journal of Optoelectronics·Laser
基金
国家自然科学基金资助项目(60673153
60970105)
山东省自然基金资助项目(Y2007G22)
关键词
μm级颗粒检测
轨迹关联
机器视觉
micro-scale particle detection
track association
machine vision