图像处理在药片视觉检测系统中的应用

摘要

Visual inspection system was used in the assembly line of tablets automatic packing.Tablets image enhancement , segmentation,and feature extraction were used in the image proceasing unit to realize the functions of packaging tablets discrimination,such as lack and defects.Original images were obtained from the camera with CCD sensor.Non-uniform illumination causes low contrast.Aimed at the problem,this paper proposed a grey scale compensation method based on Gaussian low-pass filter model.Then it segmented the images by the threshold curved surface.It can improve the performances of pixel misclassification rate and shape and make progress in accuracy.In geometry features extraction,it proposes a kind of improved perimeter calculation method,which accelerates real-time feature.%视觉检测系统可应用于药片自动包装流水线上,其中的图像处理模块经过图像增强、分割、特征提取步骤,最后实现药片包装的缺粒、缺损识别的目的.原始图像由带有CCD传感器的摄像头获取,由于光照不均匀导致图像对比度低.针对此问题,提出了基于高斯低通滤波器模型的灰度补偿方法,然后用阈值曲面去分割图像.结果在像素错分率和形状度这两项性能指标上有一定改进,提高了识别的准确率.在几何特征提取中,提出了一种改进的周长计算方法,加快了检测的实时性.