数学形态学在作物病害图像预处理中的应用研究

摘要

This paper select the images of wheat leaf disease as research object which were collected under complex back -ground , In order to analyze the type and degree of wheat diseases furtherly , an image preprocessing method based on mathematical morphology was designed according to geometry characteristic of wheat leaf .Using phase consistency model for edge detection firstly , then the closing operation , opening operation , shape area filling operation and reconstruction in combination with area-open operation are used to preprocess the original image under complex background with a lot of noise .Experimental results demonstrate that the pre-processing method proposed in this paper is efficient in a single wheat leaf extraction and noise elimination while retaining lots of detailed information of the wheat leaf original image . The method detected the prominent leaf accurately , which provides a new way for feature extraction and classification of crop disease .%　　以大田复杂背景下采集的小麦叶部病害图像为研究对象，为进一步分析小麦病害类别及程度，针对小麦叶片的几何形状特点，设计了一种基于数学形态学的小麦叶部图像预处理方法。该方法首先采用相位一致性模型进行边缘检测，然后将数学形态学中的闭运算、开运算、形态区域填充运算及形态面积开运算相结合，用于对复杂背景下混杂大量噪声的小麦叶部病害图像进行去噪、提取及重建。实验证明，该方法能有效地将图像中最突出的小麦单个叶片从复杂背景中提取出来，并能够保留原始病害图像中的病害细节信息，图像清晰完整，为作物病害的特征提取及分类识别研究提供了新的思路。