基于颜色特征的加工番茄叶片氮素评价初步研究

A Primary Study on N Evaluating of Processing Tomato Leaves Based on Color Features

利用数码相机获取加工番茄地上部分彩色图像,通过图像处理软件提取RGB及其组合的颜色特征值,同时与氮素指标叶片叶绿素含量、SPAD值、叶片含氮量及单株吸氮量作回归分析,根据统计性检验,有相当数量的颜色特征与4个氮素指标呈较高的相关性,相关系数达到r=0.7以上。综合筛选出的颜色特征指标,结合地面覆盖度与氮素营养指标建立以叶片含氮量为应变量的估算模型。经模型校验,其预测值与实测值在n=15,P<0.01水平上达r=0.8以上的极显著相关。因此,可依据颜色特征参数通过建立相应的统计模型进行加工番茄叶片氮素含量的评价,进而为加工番茄氮素营养诊断提供依据。

This research introduced CCD digital camera as a sensor through which processing tomato plant color images were captured. Then, tomato canopy and background pixel was identified by image processing software. Background noise was eliminated such as hotspot and shadow and soil etc. Color characteristics of leaf images such as R G/3 were recorded with the image processing software. Meanwhile, leaves＇ chlorophyll contents and SPAD value and leaves＂ nitrogen concentration and N absorbed content was measured in Lab. Canopy parameters and color information were paired for regression analysis. Empirical statistical showed that a significant high coefficient between the color characteristics and the four nitrogen nutrition parameters. The correlative coefficient was higher than 0.7. Furthermore, a statistic correlation models were developed between color characteristic parameters and PGCV% and nitrogen nutrition parameters. Validation test showed the correlative coefficient between the estimated value and testing value was over 0.8. Therefore, the color characteristic parameters can be used to establish statistic model and evaluate processing tomato leaves＇ nitrogen nutrition states, and for the diagnosis of N status of processing tomato.