葡萄糖注射液中不溶性微粒的识别算法研究

针对传统人工目测法以及光阻检测法在葡萄糖注射液中不溶性微粒识别判断效率较低现象,提出了一种背景差预处理与改进的Mean Shift算法相结合的微粒识别算法。利用高分辨率数码相机采集在一定照度下葡萄糖注射液中不溶性微粒产生散射光的序列数字图像。对序列数字图像进行预处理及并标定数字图像有效检测区域。建立序列数字图像背景模型并对背景噪声与随机噪声进行抑制处理。采用改进的Mean Shift算法对葡萄糖注射液中不溶性微粒进行目标跟踪。通过对序列数字图像中被检测对象的分割与特征提取,实现了葡萄糖注射液中粒径在25μm以上不溶性微粒的识别。

Study on identification algorithm of insoluble particles in glucose injection

Aiming at improving the low testing efficiency of traditional manual ocular estimation method and photoresistance detection method in identification of insoluble particles in glucose injection, a particle identification algorithm based on the combination of background subtraction pretreatment and improved mean shift algorithm is proposed.Sequential digital images of insoluble particles in glucose injection scattering lights under a certain illumination are acquired with a digital camera with high image resolution.The sequential digital images are pretreated well enough and valid surveyed area in every digital image is clearly demarcated.A model for the digital image background is built for background noises and random noises being suppressed.Insoluble particles in glucose injection are tracked with the improved mean shift algorithm.As a research result, identification of insoluble particles in glucose injection with diameter larger than 25 μm are successfully realized by division and feature extraction made to the objects to be detected in the sequential digital images.