中文摘要：

目前，脱水姜片的分选主要靠人工完成，分选精度不高，结果不稳定。为此，研究设计了一套基于计算机视觉的在线分选装置；开发了基于物料颜色参数的实时算法，对物料图像进行特征提取和一系列处理；运用双缓冲技术提高系统运行速度。试验结果表明，分选出的成品中含有的合格姜片达到95．5％，系统测量速度快、分选精度高且运行稳定，满足在线实时分选要求。

英文摘要：

At present, the detection of the dehydrated ginger on the domestic market mainly depends on artificial sorting with low accuracy. The computer vision was proposed to apply to the dehydrated ginger and a set of device for online classification has been developed. A series of real-line algorithms based on the material color parameters and double-buffering with computer technology is designed. The pretreatments indicate that the average correct classification rate is 95.5%. The results show that the system runs reliable with high measuring speed and accuracy, and it has reached the requests of online real-line classification.