基于图像识别技术的玉米种子形态鉴别研究  

Research on the Morphology Identification of Corn Seed Based on Image Recognition Technology

作　　者：李少敏[1] 吴忠秀[1] 陈升晖[1] Li Shaomin;Wu Zhongxiu;Chen Shenghui(Hainan College of Vocation and Technique,Haikou,570216)

机构地区：[1]海南职业技术学院,海口570216

出　　处：《分子植物育种》2022年第2期667-671,共5页Molecular Plant Breeding

基　　金：海南自贸港南繁科技城及国家南繁基地联合征文并由“种子特性快检技术专项目”(SSISCS2021001)资助。

摘　　要：玉米作为中国主要粮食作物之一,对于农业发展占据着尤为重要的地位,玉米种子质量的好坏直接影响到了玉米产量和品质,对于玉米种子质量的把控也成为了近年来研究的热点。随着计算机技术的不断发展,实现农作物检测自动化也成为了必然趋势,计算机视觉技术的发展凭借着无损检测、高效、高精度等特点开始广泛被应用于种子检测这一领域。本研究收集‘京科389’玉米种子图像,通过中值滤波和灰度处理对种子图像进行降噪增强,随后对该图像进行阈值分割,利用玉米种子周长、面积、圆形度参数对种子进行划分,对于玉米种子形态识别率多数超过了90%,具有较高的正确率,为种子检测和管控种子质量方面提供了研究基础。As one of the main food crops in China, maize occupies a particularly important position for agricultural development. The quality of maize seeds directly affects the yield and quality of maize. The control of maize seed quality has also become a research hotspot in recent years. With the continuous development of computer technology, the realization of crop inspection automation has also become an inevitable trend. The development of computer vision technology has begun to be widely used in the field of seed inspection due to the characteristics of non-destructive inspection, high efficiency, and high precision. ’Jingke389’ corn seed image is collected, the seed image is denoised and enhanced by median filtering and grayscale processing, and then the image is thresholded,and the corn seed perimeter, area, and circularity parameters are used to divide the seeds. The recognition rate of corn seed morphology is mostly more than 90%, with a high accuracy rate, which provides a research foundation for seed detection and control of seed quality.

关 键 词：玉米种子 计算机视觉 种子形态 图像处理 检测 

分 类 号：S513[农业科学—作物学]