香蕉叶面积回归模型的建立

Regression model building on leaf area of banana

以粉蕉、红蕉、泰国蕉、大蕉等7个香蕉品种叶片为试材,应用数字图像处理技术精确测算叶面积,探讨叶长、叶宽及叶长×叶宽与实测叶面积之间的关系,并建立普适性应用模型。结果表明,7个香蕉品种的叶面积与叶长、叶宽的回归模型均达到极显著水平,可用于香蕉叶面积估算。进一步比较各模型的决定系数(R2)和离回归平方和(Q)得知,以多项式方程和一元线性方程模拟计算的香蕉叶面积精确度更高,其中粉蕉、8818、巴西蕉建议使用拟合多项式方程来计算;红蕉、泰国蕉、皇帝蕉、大蕉建议使用拟合线性方程来计算。

Seven banana varieties including M.AAB Group Fenjiao, M.AAA Group Red green, Thailand banana, Huangdi banana （Musa paradisica AA）, 8818, Plantain and Brazil banana were used as materials, and leaf area of banana were calculated by digital image technology to study the relationships between leaf area and leaf length, leaf width, leaf lengthxleaf width and build a common application model. The results showed that the regression model between leaf area and leaf length, leaf width of seven banana varieties reached extremely significant level and could be used for estimating leaf area of banana. From the comparisons of the coefficient of determination（R^2） and square sum（Q） of partial regression, the accuracy of leaf area calculated by polynomial equation and monadic linear equation were higher, and the leaf area of M.AAB Group Fenjiao and Brazil banana could be calculated by fitting polynomial equation, that of M.AAA Group Red green, Thailand banana, Huangdi banana （Musa paradisica AA） and Plantain could be calculated by fitting linear equation.