Abstract:

To achieve high-throughput measurement of maize kernel phenotypic traits, a maize kernel traits system was designed based on line-scan technology combined with automatic control technology. The maize kernels were fed successively with vibrating feeder, and the undistorted images of corn kernels were acquired by matching the conveyor speed and line-scan speed via servo drive technology. Through image processing technique, the maize kernel phynotypic traits parameter measurement were realized. The experimental results showed that the measurement accuracy (mean absolute percentage error, MAPE) of total kernel number, long axis, short axis, and length-width ratio were 0.50%, 1.22%, 3.34%, and 4.22%, respectively. The measurement efficiency of the system was 12 s per ear. This study is of great significance for promoting the high-throughput identification of maize kernel traits.

Key words: maize kernel, line-scan, trait measurement, high-throughput

摘要：

为实现玉米籽粒表型性状参数的高通量测量，设计了一种基于线扫描技术和自动化控制技术相结合的玉米籽粒考种装置。该系统通过振动给料机实现玉米籽粒快速喂料，应用伺服驱动技术实现输送带运行速度和线阵扫描速度无偏差匹配，实现玉米籽粒图像无畸变获取，通过图像处理技术实现玉米籽粒表型性状参数的测量。试验结果表明，与人工测量值比较，该装置对玉米籽粒总粒数、长轴、短轴、长宽比的平均相对误差分别为0.50%、1.22%、3.34%、4.22%，平均测量效率为12 s/穗。该研究对于推动玉米籽粒高通量表型鉴定具有重要意义。

关键词: 玉米籽粒；线扫描；性状测量；高通量