摘要
为开发具有较高效率和精度的精密排种器试验检测系统,研究了一种机器视觉检测技术,它采用视觉传感器采集排种口动态种子流下落时的序列图像,通过图像处理、排种时间检测、型孔种子数识别等过程,获得型孔种子数的频率分布,进而计算合格指数、重播指数、漏播指数等排种质量指标。研究结果表明,采用平稳随机过程模型描述排种时间和型孔种子数序列,采用帧间隔及最大帧种子时间与排种间隔合理匹配的图像采集模式以使种子重复成像,既可保留有用信息,又可在较宽范围内选用帧频率,能够在25~30Hz常见帧频率下对单粒播和穴播排种器的排种质量进行有效检测。
A detection technique based on machine vision method was investigated, by which the higher efficiency and precision system for experimentation and detection of metering mechanism of precision drills was developed. This technique was composed of series images collected by machine vision sensor aiming at dynamic seeding flow from seeding hole, image processing, seeding time detection, cell seeding amount identification, quality of feed index and multiple index and miss index calculation. The results showed that seeding quality of metering mechanism was effectively detected in single seed drilling and hill--drop drilling at 25 -30 frame frequency. Steady random process theory was used to describe series of seeding time and series of cell seeding amount. Interval between abutting frames and time through frame altitude must matched seeding time space of metering mechanism. This technique can effectively reserve seeding information and use more frame frequency in detection.
出处
《山西农业大学学报(自然科学版)》
CAS
2008年第3期324-328,共5页
Journal of Shanxi Agricultural University(Natural Science Edition)
基金
山西高校科技研究开发项目(20051221)
山西省科技攻关项目(041085)
关键词
排种器
排种质量
检测
机器视觉
Metering mechanism
Seeding quality
Detection
Machine vision