重叠条件下茶叶嫩芽的自动检测方法

茶叶嫩芽的自动检测是实现茶叶自动化采摘的难点之一,本文针对陕西名茶“午子仙毫”茶叶图像,采用提取绿色分量、区域标记和逐行扫描相结合的方法,实现了叶片重叠条件下的嫩芽自动检测。实验中,首先提取茶叶原始彩色图像中的绿色分量,并通过阈值分割的方法,将嫩叶和老叶分离 其次采用区域标识的方法,去掉嫩叶图像中孤立噪声区域,实现茶叶“两瓣一心”区域的提取,并采用数学形态学的膨胀运算,填充图像中由于病害等引起的叶片表面缺陷 最后提出始于顶端、逐行扫描的方法,直至检测出叶片分叉部位,实现嫩芽目标的检测。实验结果表明,该方法能对30幅图像正确检测出28幅,准确率为93.3%,表明该方法可以对有叶片重叠的单株茶叶嫩芽目标进行检测。

Automatic detection of overlapped tea leaf sprouts

Detection of the tea leaf sprout is vital in the picking of tea leaves. This paper presents a meth- od used to automatically recognize the tea sprout of Shaanxi famous tea ＂ Wuzi Xianhao＂. Firstly, the green component of the tea image was extracted, and two thresholds were set to segment the old and young leaves. Secondly, a region marking method was applied to mark the region of new leaves image, and se- lect the region corresponding to the most pixels of the same label. The ＂Two leaves and a sprout＂ region could be selected by the above mentioned process. Then a dilation algorithm was used to remove the noise points. Thirdly, scan the image row by row from the first pixel until find the bifurcate point. Finally the sprout could be detected. The experiment results showed that 28 out of 30 images were recognized correct- ly with an accurate rate of 93.3% , and indicate that this image processing method can be used in the detection of overlapped tea leaf sprouts.