摘要
运用机器视觉技术实现条播排种器性能的自动化检测。研究表明,帧种子数分布相对于拍摄频率(帧频率)具有独立性。以该独立性为基础构造的机器视觉检测方法为:采用CCD摄像机直接拍摄条播排种器试验的动态种子流,对获得的样本进行图像标定、灰度化处理、非种子目标去除、二值化处理、滤波去噪、小目标去除、种子遮挡判别、排种模式识别等图像处理过程,估计排种量的分布和计算排种器性能指标。
The machine vision method was adopted to realize automatic detection of performance for metering mechanism of a drill. The external force feed test indicated that the frequency of seed amount in frame is independent of shooting frequency. On the basis of this independence the methods of machine recognizable detecting were formed. A dynamic seeding flow in metering mechanism experiment of the drill was shot by a CCD camera and a＇number of steps were taken including image standardization, gray treatment, the deleting of targets without seeds, binary number treatment, noise-removing of filter, the abandon of mini-targets, identification of seed cover, and metering model. Finally, metering amount distribution was estimated and each index of metering mechanism performance was worked out.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2005年第11期50-54,49,共6页
Transactions of the Chinese Society for Agricultural Machinery
基金
山西省科技攻关项目(项目编号:041085)
关键词
排种器
性能
机器视觉
检测
Metering mechanism, Performance, Machine vision, Detection
