机器视觉技术对氯化三苯基四氮唑染色法鉴定玉米种子生活力的改进

Improvement of machine vision technology for determination of maize seed viability by tetrazole staining

为探讨机器视觉技术在提高玉米种子生活力检测结果判定上的准确性和可重复性,以‘三北6号’玉米种子为试验材料,将吸胀的玉米种子沿胚部纵切,置于0.1%氯化三苯基四氮唑(TTC)溶液,30℃黑暗条件下染色2h,使用Image J和PhenoSeed软件进行图像分析,结合人工观察鉴定,分析玉米种子胚部的颜色指标、染色比例与种子生活力之间的关系。结果表明,基于机器视觉技术提取的‘三北6号’玉米种子TTC染色后胚部的颜色特征建立的生活力检测模型,准确率均达到83%以上。随机选取6份其他品种玉米种子进行模型验证,结果表明发芽率的预测值与真实值之间无显著差异。

In order to explore the accuracy and repeatability of machine vision technology in determination of maize seed viability,the seeds of‘Sanbei 6’maize variety were taken as experimental materials in this study.The swelling maize seeds were cut longitudinally along the embryo,and then placed in 0.1% triphenyl tetrazolium chloride(TTC)solution,and stained for 2hat 30℃ in dark.Image J and PhenoSeed software were used for image analysis.Combined with artificial observation and identification,the relationship between the color index of maize seed embryo,the staining ratio and seed viability was analyzed.The results showed that the accuracy rate of vitality detection model based on color characteristics of embryo after TTC staining of‘Sanbei 6’maize seeds is more than 83%.The seeds of six other maize varieties were randomly selected for model validation.The results showed that there was no significant difference between the predicted germination percentage and the true value.