磁吸式精密播种器性能图像检测技术研究

STUDY ON IMAGE DETECTION TECHNOLOGY FOR PERFORMANCE OF MAGNETIC TYPE PRECISION SEEDER

根据磁吸式精密播种器结构及工作特点,运用机器视觉技术,建立排种性能视觉检测系统。采用摄像头与磁吸头相对静止,伴随排种机构运动的摄取方式,避免图像运动失真;采取图像平滑、分割等预处理方法,有效地提高图像质量;提出基于形态学图像处理的种子特征提取法,通过种子特征判断磁吸头取种情况,由此统计播种器在某一时间段的播种精度,包括单粒率、重播率、漏播率。试验表明,采用该系统进行实时检测,效率高,其漏播识别准确率高达100%、单粒识别准确率达95%以上。

According to the structure and working characters of magnetic type precision seeder, an image detection system for seed - metering performance was established by means of computer vision technique. The system adopts the shooting method of keeping CCD camera relatively static with the electromagnetic sucker and moving together with seed-metering mechanism when collecting the image, the motion distortion of images is avoided. By using image pre-treatment techniques of smoothing, segmenting, etc. , image quality was enhanced effectively. And the technology of seed features extraction based on morphologic image processing is put forward, it will estimate the seeds picking status of the magnetic sucker through the feature of seeds, then makes statistic of sowing precision of the seeder during a certain period of time, including single grain rate, missing rate and over-sewing rate. The experiments show that this system is of high efficiency when used in real-time detection, the accuracy of missing rate identification and single grain rate identification are up to 100% and more than 95% respectively.