论文部分内容阅读
为了快速、无损地获得苹果叶片叶绿素含量与其表面颜色特征之间的关系,为诊断苹果树生理状况提供科学依据。以新梢旺长期的红富士苹果树为研究对象,应用数码相机采集叶片图像,利用图像处理技术,采集叶片图像的红(R)、绿(G)和蓝(B)值,通过运算组合构造颜色特征参数,建立基于苹果叶片颜色特征参数的叶绿素含量估算模型,并对其精度进行评价和验证。结果表明,叶绿素含量敏感的颜色参数分别为B、B/R、B/G、G/(R+G+B)、B/(R+G+B)、(R–B)/(R+B)、(G–B)/(G+B)、(R–B)/(R+G+B)和(G–B)/(R+G+B)值;基于以上9个敏感颜色参数分别建立单变量回归模型和支持向量机回归模型(SVM),估测叶片Chl.a、Chl.b、Chl.(a+b)和SPAD值,其中单变量回归模型决定系数(R~2)均在0.6左右;SVM回归模型的决定系数(R~2)分别为0.8754、0.8374、0.8671和0.8129,均方根误差(RMSE)分别为0.0194、0.0350、0.0497和0.9281,相对误差(RE)分别为0.8059%、1.7540%、1.1224%和1.1894%,尤以对Chl.a的估测效果最佳,SVM的估测精度高于单变量回归模型。模型验证取自1/4同样本数据,验证结果表明基于SVM的Chl.a稳定性更佳,R~2=0.8275,RMSE=0.0293,RE=1.8529%。应用数码相机并基于RGB颜色模型可快速估测苹果叶片叶绿素含量,可对果园水肥的精确管理提供技术支持。
In order to quickly and non-destructively obtain the relationship between the chlorophyll content of apple leaves and its surface color characteristics, provide a scientific basis for the diagnosis of apple tree physiological status. Taking the red Fuji apple tree of Shouwangwang as the research object, the digital camera was used to collect the leaf images, and the red (R), green (G) and blue (B) values of the leaf images were collected by using the image processing technology. Color characteristic parameters, the chlorophyll content estimation model based on the color parameters of apple leaves was established, and its accuracy was evaluated and verified. The results showed that the color parameters sensitive to chlorophyll content were B, B / R, B / G, G / R + G + B, B / R + G + B, (G-B) / (G + B), (R-B) / (R + G + B) and (G-B) / (R + G + B) values based on the above 9 sensitive colors The parameters Chl.a, Chl.b, Chl. (A + b) and SPAD were estimated by using univariate regression model and support vector regression model (SVM) respectively. The univariate regression model was used to determine the coefficients ) Were all around 0.6. The coefficient of determination (R ~ 2) of SVM regression models were 0.8754, 0.8374, 0.8671 and 0.8129 respectively, and the root mean square error (RMSE) were 0.0194, 0.0350, 0.0497 and 0.9281, Which is 0.8059%, 1.7540%, 1.1224% and 1.1894% respectively, especially for Chl.a, and SVM has higher estimation accuracy than univariate regression model. The model verification is taken from the same data of 1/4 and the verification results show that the stability of Chl.a based on SVM is better. R ~ 2 = 0.8275, RMSE = 0.0293, RE = 1.8529%. Applying a digital camera and estimating the chlorophyll content of apple leaves rapidly based on the RGB color model can provide technical support for the accurate management of water and fertilizer in the orchard.