基于计算机视觉的田间杂草识别方法研究

Research on the Method of Identifying Field Weed by Computer Vision

杂草对农作物的侵害一直是人们关注的问题。为此,利用计算机图像处理技术对复杂土壤背景下的田间杂草进行识别。研究中采用可变阈值法滤除了土壤背景,应用形态学方法识别出了杂草。同时,根据杂草投影中心的像素位置确定田间杂草的实际位置。实验结果表明,此方法可有效地从阔叶子植物中识别出窄叶子稗草。

The weed to crops＇ violation that people always concern .The technology of image manipulation by using computer helps to identify the field weed in under the complex soil background conditions. Changing threshold segmentation was applied to filtrate conditions and morphological algorithm to identify the weed. The actual position of the field weed can be ascertained according to the pelts position of the weed＇s projective center. The experimental results revealed that this method can be devoted to distinguish the narrow leaf barnyard grass weeds from the extravagant leaf plant.