摘要
为解决谷物外观品质检测过程中获取图像的籽粒粘连问题 ,提出了一种基于先验知识的流域分割算法。首先 ,设定合理的区域面积阈值 Ath,标记并去除图像中的单个籽粒区域 ,结果记为 Ms;然后 ,设定有效腐蚀次数 Eth,对粘连籽粒区域进行流域分割 ,结果记为 Mc;最后 ,取 Ms与 Mc的并集 ,作为分割结果。试验结果表明 ,该算法分割效果较好 ,分割精度达到 95 .4 %。
A new watershed algorithm based on priori knowledge is presented in this paper, which was adapted to paddy image segmentation. Firstly, single seed regions are marked and removed from the clustering paddy image according to single area threshold (Ath), and the result image is marked Ms. Then, the clustering seed regions are segmented by watershed algorithm according to effective erosion degree threshold (Eth), and the segmented image is marked Mc. Finally, the union of Ms and Mc is taken as the result image. The experimental results showed that the method was valid and efficient, and its precision reached 95.4%.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2005年第3期95-98,共4页
Transactions of the Chinese Society for Agricultural Machinery
基金
"九五"国家科技攻关计划资助项目 (项目编号 :990 10 0 112 )
