基于图像处理的玉米叶片含水率诊断方法研究

Research on methods of diagnosing crop water-deficiency based on image processing

研究了利用数字图像处理技术进行作物叶片含水率诊断的方法。以温室中培育的90株不同灌水量的盆栽玉米为研究对象,使用佳能IXUS110的1 210万像素数码相机采集离体抽穗期玉米叶片的图像信息,然后利用烘干法测量叶片样本的含水率;利用叶片图像的灰度直方图提取叶片图像的均值、峰态、方差、歪斜度、能量、熵六组特征值。利用提取的20组玉米叶片样本的数据,采用线性回归的方法建立均值与玉米叶片含水率之间的关系模型;使用其余20组样本对模型进行验证,其标准差为0.021。结果表明,利用作物叶片灰度直方图均值参数可以对玉米的叶片含水率进行预测。

Digital image processing techniques were used to evaluate crop water stress, by cultivating 90 plants of corn with different irrigation amounts in a greenhouse. The Canon IXUSll0 digital camera with 12.1 million pixels was used to capture the images of corn leaves after being picked from the plants during the heading stage, and then the mois- ture content of the leaves was detected by using drying method. The eigenvalues of mean, kurtosis, variance, skew de- gree, energy and entropy were calculated by using grey histogram of leaf images. The data extracted from the leaves of 20 samples were used to set up the linear regression model showing the relationship between the mean and the leaf moisture, and the other 20 samples were used to verify the model. The standard deviation of the validation results was 0. 021. It was concluded that of the eigenvalue of mean of leaf images could be used to predict the moisture content in corn leaves.