文献类型：期刊文献

中文题名：改进Otsu法的冠层图像分割及银杏叶面积指数估测

英文题名：Canopy Image Segmentation Based on Improved Otsu Method and Leaf Area Index Estimation of Ginkgo biloba

作者：李晓冬[1] 王雪峰[1] 贺鹏[1]

第一作者：李晓冬

机构：[1]中国林业科学研究院资源信息研究所

年份：2013

卷号：41

期号：8

起止页码：52-56

中文期刊名：东北林业大学学报

外文期刊名：Journal of Northeast Forestry University

收录：CSTPCD;;北大核心:【北大核心2011】；CSCD:【CSCD2013_2014】；

基金："948"项目(2011-4-67)

语种：中文

中文关键词：图像分割;Otsu法;最大类间方差;叶面积指数;模型

外文关键词：Image segmentation; Otsu method; Maximum between-class variance; Leaf area index; Model

分类号：S757.3

摘要：基于Otsu法对冠层图像临界处分割不准确的缺点,结合类间方差以及类内聚度对阈值选取及图像分割效果的影响,提出了一种改进阈值选取算法。利用银杏冠层图像分割实例进行比较,同时进一步拟合了图像信息与叶面积指数间模型,结果表明:(1)该改进法较传统Otsu法可得到更好的分割效果;(2)以分割得到的前景像素比值作为自变量,叶面积指数为因变量,拟合得到的模型能较好的描述冠层图像信息与叶面积指数间的关系;(3)提出了一种即时无损并快速可靠的叶面积指数估测方法。
Considering the poor segmentation result of the canopy images by Otsu, we improved the threshold selection algorithm with both between-class variance and within-class variance. We segmented the canopy images of Ginkgo biloba, and then fitted the model between image information by improved Otsu segmentation and the leaf area index. Better segmentation re- suits can be obtained through the improved method rather than traditional Otsu method. The fitting model can well figure out the relationship between segmented image information and Ginkgo biloba leaf area index. Therefore, we presented a non-destructive, fast and reliable estimation method of leaf area index.