基于数码图像的棉花叶片氮含量估测研究

为了探究利用数码图像处理技术诊断棉花氮素营养的最佳图像特征参数和最佳叶位,以‘新陆早53号’为材料,使用智能手机获取棉花不同叶位叶片图像,利用数字图像处理技术提取叶片图像颜色和纹理特征参数,分析叶片氮含量和特征参数的相关性,构建基于不同图像特征参数的氮含量估测模型。结果表明：颜色特征参数r、G/(B+R)、G/B和纹理特征参数COR、ASM与叶片氮含量相关性较好,相关性系数(r)均大于0.55。基于颜色-纹理综合特征参数构建的氮含量估测模型均优于基于颜色或纹理特征参数所建模型,其中倒4叶氮含量估测模型最优,模型决定系数(R²)为0.875、均方根误差(RMSE)为1.324、相对误差(RE)为8.09%。因此,利用数字图像处理进行棉花氮素营养诊断时,应选择的最佳图像特征参数为r、G/(B+R)、G/B、COR、ASM,应选择的最佳叶位为棉花倒4叶。

The Nitrogen Content in Cotton Leaves: Estimation Based on Digital Image

In order to study the best image characteristic parameters and leaf position for diagnosing cotton nitrogen nutrition by digital image processing techniques, the cotton variety ‘Xinluzao53’ was used as the material, and the leaf images in different leaf positions were obtained by smartphone. The color and texture characteristic parameters of cotton leaf were extracted by digital image processing techniques, the correlation was analyzed between nitrogen content in leaves and characteristic parameters, and the estimation model of nitrogen contents was established based on different characteristic parameters. The results showed that the color characteristic parameters r, G/(B+R) and G/B, and texture characteristic parameters COR and ASM had good correlation with leaf nitrogen content, and the correlation coefficients (r) were all greater than 0.55. The estimation models of nitrogen contents in different leaf positions based on the color-texture comprehensive characteristic parameters were better than the models based on the color or texture characteristic parameters. Among them, the nitrogen content estimation model of the 4th leaf from the bottom was the best, the model determination coefficient (R ²) was 0.875, the root mean square error (RMSE) was 1.324, and the relative error (RE) was 8.09%. Therefore, when using digital image processing for cotton nitrogen nutrition diagnosis, the best image characteristic parameters should be selected as r, G/(B+R), G/B, COR, ASM, and the best leaf position should be the 4th leaf from the bottom.