基于MRFO的茶叶嫩芽图像分割方法

Tea Sprout Image Segmentation Method Based on MRFO

针对名优茶早期嫩芽检测问题,提出一种基于魔鬼鱼觅食优化(MRFO)的颜色因子与阈值的嫩芽图像分割算法。首先,利用MRFO优化算法训练得到ExG系数,用ExG对嫩芽RGB彩图进行灰度化,并归一化处理得到灰度图;然后,利用Otsu进行图像阈值计算;最后,对阈值进行偏移校正,获得最佳阈值。正视图和斜视图试验结果显示,此算法能突出嫩芽与老叶像素值的差异,很好地检测出茶叶嫩芽,Jaccard,Dice,Bfscore平均值分别为57.25%,72.35%,80.215%。

In order to detect the early buds of famous tea,an image segmentation algorithm based on the color factor and threshold of Manta ray foraging optimization(MRFO)is proposed.Firstly,the ExG coefficient is obtained by MRFO optimization algorithm training;secondly,the RGB color image of shoots is grayed by ExG,and the grayscale image is obtained by normalization processing;then,the image threshold is calculated by Otsu;finally,the threshold is offset corrected to obtain the best threshold.The experiment results of front view and slant view show that the algorithm can highlight the difference of pixel value between the young and the old leaves,and the tea shoots are detected very well.The average values of Jaccard,Dice and Bfscore are 57.25%,72.35%and 80.215%,respectively.