在绿肥翻压条件下, 常规的玉米氮素营养诊断技术存在耗时、 费力和可靠性差的缺点。 基于数码相机的可见光光谱技术已被广泛应用于大田作物的氮素营养诊断, 但尚未见应用于绿肥翻压后的玉米氮素营养诊断。 为评价利用图像处理技术进行绿肥翻压后玉米氮素营养诊断和玉米产量预测的可行性, 设置了不同施氮水平下的绿肥翻压试验, 利用数码相机获取不同生育期玉米冠层数字图像, 分析了玉米冠层图像色彩参数与氮素营养诊断指标和成熟期籽粒产量之间的关系。 结果表明, 绿肥翻压显著改善了玉米的氮素营养, 不同生育期的玉米叶绿素含量(SPAD值)、 地上部生物量和吸氮量均高于单施化肥处理; 绿肥翻压处理下, 玉米冠层光谱指数与氮素营养指标间的相关性较单施化肥处理低, 且其相关性在不同的生育期有较大变异, 其中, 12叶期(V12)的蓝光标准化值(B/(R+G+B))与灌浆期(R4)的红光标准化值(R/(R+G+B))与植株氮营养指标相关性较好, 二者均与玉米产量间呈显著直线回归关系, 回归系数分别为45%和46%。 因此, 数字图像技术在进行绿肥翻压后玉米氮素营养的诊断和产量预测方面具有应用潜力, 但应注意诊断时期和关键指标的选择。

In order to explore the feasibility of using the image processing technology to diagnose the nitrogen status and to predict the maize yield, a field experiment with different nitrogen rates with green manure incorporation was conducted. Maize canopy digital images over a range of growth stages were captured by digital camera. Maize nitrogen status and the relationships between image color indices derived by digital camera for maize at different growth stages and maize nitrogen status indicators were analyzed. These digital camera sourced image color indices at different growth stages for maize were also regressed with maize grain yield at maturity. The results showed that the plant nitrogen status for maize was improved by green manure application. The leaf chlorophyll content (SPAD value), aboveground biomass and nitrogen uptake for green manure treatments at different maize growth stages were all higher than that for chemical fertilization treatments. The correlations between spectral indices with plant nitrogen indicators for maize affected by green manure application were weaker than that affected by chemical fertilization. And the correlation coefficients for green manure application were ranged with the maize growth stages changes. The best spectral indices for diagnosis of plant nitrogen status after green manure incorporation were normalized blue value (B/(R+G+B)) at 12-leaf (V12) stage and normalized red value (R/(R+G+B)) at grain-filling (R4) stage individually. The coefficients of determination based on linear regression were 0.45 and 0.46 for B/(R+G+B) at V12 stage and R/(R+G+B) at R4 stage respectively, acting as a predictor of maize yield response to nitrogen affected by green manure incorporation. Our findings suggested that digital image technique could be a potential tool for in-season prediction of the nitrogen status and grain yield for maize after green manure incorporation when the suitable growth stages and spectral indices for diagnosis were selected.