摘要
针对采茶机器人的茶叶嫩芽识别问题,提出一种基于颜色因子与图像融合的茶叶嫩芽图像检测算法。首先对RGB彩图进行ExG、ExG-ExR,MExG,COM2灰度化处理,并进行灰度图归一化处理;然后选择合适的通道,利用Haar和DB2小波进行多通道图像分解、滤波、融合。获得融合后的灰度图像直方图,对直方图形状进行分析,根据嫩芽老叶的面积比与像素数目比确定图像分割阈值。试验结果表明,此算法能充分利用嫩芽与老叶的颜色差异,很好地检测出茶叶嫩芽,SD,Dice,ER,NR分别为63.005%,60.09%,101.235%,6.515%,性能优于Otsu。
Aiming at the problem of tea bud identification in tea picking robots,an image detection algorithm based on color factor and image fusion is proposed.First,the RGB color map is subjected to Grayscale processing of ExG,ExG-ExR,MExG,and COM2,and the grayscale normalization processing is performed;then the appropriate channel is selected,and the multi-channel image decomposition,filtering,and fusion are performed by using Haar and DB2 wavelets.The merged gray image histogram is obtained,the histogram shape is analyzed,and the image segmentation threshold is determined according to the area ratio of the old shoots and the ratio of the number of pixels.The experimental results show that the algorithm can make full use of the color difference between the shoots and the old leaves,and detect the tea shoots well.The SD,Dice,ER and NR are 83.41%,90.96%,16.97%and 2.43%respectively,and the performance is better than Otsu.
作者
姜苗苗
问美倩
周宇
杨芷羽
王铂闻
程玉柱
Jiang Miaomiao;Wen Meiqian;Zhou Yu;Yang Zhiyu;Wang Bowen;Cheng Yuzhu(School of Mechanical and Electronic Engineering,Nanjing Forestry University,Nanjing City,Jiangsu Province 210037,China)
出处
《农业装备与车辆工程》
2020年第10期44-47,共4页
Agricultural Equipment & Vehicle Engineering
基金
南京林业大学大学生创新项目(2019NFUSPITP0172)。
