种子包衣合格率图像检测系统设计与试验

Design and test of image detection system for qualification rate of seed coating

为满足对种子包衣合格率定量检测的需求,设计一套基于图像处理的检测系统,由上料下料系统、图像采集系统、控制系统和壳体等部件组成。该系统通过4个工业相机获取种子表面图像,通过工控机和软件进行处理,分别计算种子表面积和包衣面积,得到每粒种子包衣覆盖率和一批种子的包衣合格率。试验结果表明,单粒种子的平均检测时间小于3 s,覆盖率准确度的检测误差小于0.5%,覆盖率一致性偏差小于4%;一批种子合格率重复率为98.12%,差错率小于1%。与传统检测系统相比,该系统检测速度快,检测精度较高,差错率小,结构紧凑,能满足种子包衣合格率的定量检测。

In order to meet the requirement of quantitative detection of coating qualification rate, a set of detection system based on image processing is designed, which consists of feeding and unloading system, image acquisition system, control system and shell. The system obtains the image of seed surface by four industrial cameras, processes it by industrial computer and software, calculates the seed surface area and coating area respectively, and obtains the coverage rate of each seed and the qualified rate of a batch of seeds. The results showed that the average detection time of single seed was less than 3 seconds, the detection error of coverage accuracy was less than 0.5%, the consistency error of coverage was less than 4%, the repetition rate of qualified rate of a batch of seeds was 98.12%, and the error rate was less than 1%. Compared with the traditional detection system, the system has fast detection speed, high detection accuracy, low error rate, compact structure, and can meet the quantitative detection of the qualified rate of coating.