摘要
为了实现温室植物病害的智能化防治,采用数字图像处理技术对植物病害进行诊断已成为主要技术之一。就黄瓜侵染性病害的图象处理及特征值提取方法进行了研究。研究在图像预处理中,由于病变叶片形状大小不一、背景的存在和病状的不同,必然套对图像处理造成影响。本研究以选择图像的背景为切入点,分离了叶片与背景；利用边框裁减算法进行了图像处理窗口的确定,从而有效地提高了处理速度；比较了两种滤波的效果：选择邻域均值法作为颜色特征提取前的预处理,选择中值滤波作为纹理特征提取前的预处理；根据RGB三体对人眼的刺激程度的不同对图像进行了灰度化处理,从而简化了图像的处理难度．并提高了处理速度；利用双峰法从背景中分割出有效分析叶片部分,利用OSTU法将叶片正常部位与病态部位分割出来。从而为后续提取有效的病害诊断特征参数和建立模式识剐系统奠定了基础。
In order to realize the intellectual control of the greenhouse plant disease , the digital image processing technology has already become one of the main technologies for to diagnose them. This paper study the method of image processing and extracting of the characteristic values for greenhouse growing cucumber disease. In the image pre-processing, because of the different sizes of pathological leaves, and the background and the different disease, the effect of the image processing may be affected. At first, leaves are separated from background, then it sets up the image processing windows by the frame to cut down algorithms, thus doing so can improve processing speed; It also compares the results of two different kinds of flatting measures to choose neighborhood average pre-processing for drawing color feature parameters and median filtering pre-proceasing for texture feature parameters; Then according to incitement degree of RGB to eyes, the images are ashy and simplify in the imaging processing to improve processing speed. Finally, it utilizes the bimodal law to cut apart the leaves from the background and by OSTU law to cut apart normal part from disease part of leaves. Thus the image pre-processing made foundation for following effective characteristic parameters for the disease diagnoses and for setting up mode recognition system.
出处
《山西农业大学学报（自然科学版）》
CAS
2006年第4期351-354,共4页
Journal of Shanxi Agricultural University(Natural Science Edition)
基金
山西省科技攻关项目(20041018)
关键词
植物病害
图像处理
图像预处理
Plant disease Image processing Image pre-processing
作者简介
刘志华（1981-），女（汉），山西右玉人，硕士，主要从事农业生物环境与能源工程方面的研究
