摘要
变量喷雾是现代农药精确喷施技术之一。为了实现果树变量喷雾,需要检测果树的形态结构参数。图像处理技术是获取果树形态结构参数的重要方法。本文根据果树图像的特征,分别基于灰度模型、RGB模型和HSV模型对果树图像进行图像分割试验。试验结果表明,有效的果树图像分割方法是基于灰度模型的图像分割与基于RGB模型的(R+G+B)、2R、2G等分量的图像分割。其中,基于灰度模型的图像分割采用固定阈值128,基于RGB模型的图像分割采用Otsu算法。比较各种有效图像分割方法的准确率和耗时,结果表明采用基于RGB模型的2R分量、Otsu算法可以满足果树图像分割处理的要求,该方法分割准确率为79.67%。
Variable-rate spray is a modern pesticide precision spraying technology. It is necessary to detect the fruiter shape structure parameters for fruiter variable-rate spray, and image processing technique is an important method to realize it. According to the character of fruiter image, the image segmentation was experimented based on the gray model, RGB model and HSV model respectively. The effective fruiter image segmentation methods are gray model and RGB model which based on R＋G＋B, 2R and 2G component. Gray image segmentation used fixed threshold 128. RGB image segmentation algorithm was Otsu. The ac-curate ratio and time consume were compared within effective image segmentation algorithms. The result indicated that utilizing RGB based on 2R component and Otsu algorithm could meet the requirement of fruiter image segmentation with the 79.67% accurate rate.
出处
《中国农机化学报》
北大核心
2014年第5期81-86,共6页
Journal of Chinese Agricultural Mechanization
基金
国家自然科学基金资助项目(31060171)
