基于数字图像分析技术的橡胶树叶片氮含量预测  

Study on Predicting Nitrogen Content of Rubber Tree Leaf by Digital Image Analysis

作　　者：张培松[1] 孙毅明[1] 郭澎涛[1] 袁忠志 杨红竹[1] 贝美容[1] 罗微[1] 

机构地区：[1]中国热带农业科学院橡胶研究所,海南儋州571737 [2]海南晓晨科技有限公司,海南海口570125

出　　处：《热带作物学报》2015年第12期2120-2124,共5页Chinese Journal of Tropical Crops

基　　金：海南省自然科学基金(No.310097);中国热带农业科学院橡胶研究所基本科研业务费专项(No.1630022011020);中国热带农业科学院院本级基本科研业务费专项资金(No.163002214022);海南省重点科技计划项目(No.ZDXM20130045)

摘　　要：为建立橡胶树氮素营养快速诊断技术,利用数码相机获取橡胶树叶片图像,运用数字图像处理技术提取叶片图像的颜色特征参数,分析颜色特征参数与橡胶树叶片氮含量的相关性,并建立回归模型。结果表明,9个颜色特征参数R/B、B/(R+G)、R/(G+B)、R/(R+G+B)、B/(R+G+B)、(B+G)/(R+G+B)、(R-B)/(R+G+B)、(R-B)/(B+R)和G/(R-B)与橡胶树叶片氮含量相关性较好,综合评价得出G/(R-B)所建立的橡胶树热研7-33-97叶片氮含量二次多项式估测模型最优,模型校正决定系数为81.04%,预测相对误差和均方根误差分别为10.91%和0.31%,表明利用数字图像分析技术可以进行成龄橡胶树热研7-33-97叶片氮素含量营养诊断。Predicting nitrogen content of leaf based on digital image analysis, is one of the rapid diagnostic technology of rubber tree nitrogen content. This study extracted the color feature of rubber tree leaf by digital image processing techniques, which were acquired by digital camera. The regression models were estabilished base on analysised the correlation between these parameters with rubber tree leaf nitrogen content. The result showed that, the 9 color feature parameters, R/B, B/（R＋G）, R/（G＋B）, R/（R＋G＋B）, B/（R＋G＋B）, （B＋G）/（R＋G＋B）, （R-B）/（R＋G＋B）, （R-B）/（B＋R） and G/（R-B）, had good correlation with the nitrogen content of rubber tree leaf. The model based on G/（R-B） was the best one of the 9 parameters for predicting nitrogen content of reyan 7-33-97, which the correction coefficient was 81.04%, the relative error and the root mean square error were 10.91% and 0.31%, respectively, which indicated that the digital image analysis technology could be used for diagnosis nitrogen content of aged rubber tree leaves reyan 7-33-97.

关 键 词：数码相机 图像识别 橡胶树 氮素 营养诊断 

分 类 号：S794.1[农业科学—林木遗传育种]