成熟期苹果计数方法研究——以黄土高原样本为例证  

A Study on the Method of Counting Number of Ripe Apple Fruits Based on the Sample Obtained from Chinese Loess Plateau

作　　者：张万红[1] 朱元骏[1] 

机构地区：[1]西北农林科技大学水土保持研究所,陕西杨凌712100

出　　处：《西南大学学报（自然科学版）》2018年第1期27-34,共8页Journal of Southwest University(Natural Science Edition)

基　　金：中国科学院重点部署项目(KFZD-SW-306-03)

摘　　要：对比RGB(红,绿,蓝)颜色空间下各颜色分量间多种色差运算的自适应阈值分割与基于H(色调)和S(饱和度)的K均值聚类算法对苹果影像的分割,选取适宜于自然生长状态下成熟期苹果影像分割的最佳算法分割目标物.在影像分割的基础之上,通过计算影像中苹果区域的总面积与单果平均面积之商确定苹果数目.试验结果表明:1.1×R-G色差运算结合自适应阈值分割算法对成熟期苹果影像有较好的分割效果;以影像中苹果总面积与单果平均面积之商确定苹果数目的算法准确率可达82.18%,计数方法准确率高.Firstly,images of apple fruit were segmented based on the suitable image segmentation method which was generated by comparison of the adaptive thresholding segmentation of diverse color difference for RGB（red,green,blue）color model with the clustering analyzing H（hue）and S（saturation）of HIS（hue,intensity and saturation）color model.Then the number of apple fruits was determined by the quotient of total area of all apple fruits divided by average area of single apple fruit.The results showed that apple fruits could be accurately identified from images when using adaptive thresholding segmentation algorithm for 1.1×R-Gimages and this counting method gave an accuracy of 82.18 percent.It is concluded that the accuracy of the method proposed for counting number of apple fruits in this paper is high.

关 键 词：苹果影像 色差运算 自适应阈值分割 计数 

分 类 号：S66-33[农业科学—果树学]