Authors

Abstract

While studying the coating theory, due to the lack of the support of the rapid identification and detection device for coated red clover seeds, for a long time, we have mainly relied on manual visual inspection to sort qualified coated seeds, only relying on human eyes to identify the cause of low efficiency, high wrong classification rate and high labor intensity. In order to identify the coated red clover seeds quickly and efficiently, a set of intelligent identification and detection system for coated red clover seeds was designed. First of all, by building a machine vision shooting platform to ensure that the light source and other shooting conditions are consistent, the images are transmitted to Vision Assistant 2018 for image processing. Secondly, two image processing algorithms are designed to process qualified coated seeds and damaged coated seeds respectively. Finally, an identification and detection algorithm is proposed, which uses LabVIEW2018 as the host computer to identify the qualified number and the damaged number. Taking red clover seeds as the test object, the test results show that the entire system takes about 1 second to collect and process a single image; the recognition accuracy of qualified coated seeds and damaged coated seeds is above 96% and 85%. The identification and detection system realizes the nondestructive detection of coated seeds, and provides theoretical basis and technical support for the later research on the optimal seed coating process, deepening the theoretical research of the coating machine and improving the degree of automation.

Abstract in Chinese

在对包衣理论进行研究的同时，由于缺少包衣红三叶种子快速识别检测装置的支持，长期以来主要依靠人工目测分选合格的包衣种子，仅靠人眼识别效率低、错分率高、劳动强度高，为此设计了一套包衣红三叶种子智能识别检测系统，针对包衣红三叶种子进行识别。首先，通过搭建机械视觉拍摄平台，保证光源等拍摄条件一致，传输图像至Vision Assistant 2018进行图像处理。其次，设计两种图像处理算法，分别对合格包衣种子以及破损包衣种子进行处理。最后提出了一种识别检测算法,采用LabVIEW2018作为上位机对合格数以及破损数进行识别。以红三叶种子为试验对象，试验结果表明：整套系统对单幅图像采集和处理时间约为1s；对合格包衣种子以及破损包衣种子识别准确率分别在96%和85%以上。该识别检测系统实现了对包衣种子的无损检测，为后期研究种子最佳包衣工艺，深化包衣机理论研究以及提高自动化程度提供了理论基础与技术支持。