摘要
图像分割是图像预处理中的关键步骤,该步骤是后期特征提取和识别的基础。大田环境下群体小麦图像受光照和背景影响,使用经典分割算法精度较低,针对大田环境下群体小麦图像分割问题,提出了1种基于RGB颜色空间的带限定条件的GMR(RSTR-GMR)阈值分割算法,实现了小麦与背景的分离。实验结果表明:分割精确度达到82.3%,RMSE为7.2。将分割图像与HIS空间下的H分量K均值分割算法(H-K)进行了比较,在所有拍摄条件下RSTR-GMR算法均优于H-K。
Image segmentation is a key step in image preprocessing,which is the basis for later feature extraction and recognition.The image of group wheat in the field environment is affected by illumination and background,and the accuracy of using classical segmentation algorithm is low.Aiming at the problem of image segmentation of group wheat in field environment,a threshold algorithm of GMR(RSTR-GMR)based on RGB color space is proposed.The segmentation algorithm achieves the separation of wheat from the background.The experimental results show that the segmentation accuracy reaches 82.3%and the RMSE is 7.2.The segmentation image is compared with the H-component K-means segmentation algorithm(H-K)in HIS space,and the result indicates that the RSTR-GMR algorithm is better than H-K under all shooting condition.
作者
宋宇斐
滕桂法
苑迎春
刘天真
潘飞
SONG Yufei;TENG Guifa;YUAN Yingchun;LIU Tianzhen;PAN Fei(School of Information Science and Technology,Hebei Agricultural University,Baoding 071001,China)
出处
《河北农业大学学报》
CAS
CSCD
北大核心
2019年第6期128-133,共6页
Journal of Hebei Agricultural University
基金
“十三五”国家重点研发计划项目(2016YFD0200403)
河北省高等学校科学技术研究项目(ZD2015054)
河北省研究生创新资助项目(CXZZBS2017059)
河北省自然科学基金项目(G2018204093)
