模糊植物病虫害图像的检测  

Low Quality Plant Diseases and Insect Pests Image Detection

作　　者：张恒[1,2] 陈丽娟[1,2] 张岩[1] 

机构地区：[1]中国科学院新疆理化技术研究所,新疆维吾尔自治区乌鲁木齐830011 [2]中国科学院研究生院,北京100049

出　　处：《计算机仿真》2012年第1期199-201,220,共4页Computer Simulation

摘　　要：关于植物病虫害图像准确检测问题,由于通过卫星遥感技术采集的图像不清。传统的植物病虫害检测算法依靠清晰的像素信息,因采集的病虫害图像质量较低,图像模糊,造成关键像素信息的丢失、混浊,算法存在着病斑检测结果不准确的问题。为了提高准确性,提出了低质量图像的病虫害检测方法。通过建立细节点的特征库,利用模糊判别的方法对建立的少量细节特征进行检测,然后判别是否为病虫害特征.新方案只需极少的细节特征就能完成了检测,避免了对大规模像素的依赖。实验表明,采用的方法能够有效分割大部分低质量图像的病虫害特征,取得了比较好的效果。Study plant diseases and insect pests base on images. Traditional plant diseases and insect pests detection algorithm depends on clear pest pixel, when the quality of the pest images is poorer, the detection result is bad. Aiming at the problems, the paper put forward a pest detection method based on the low quality of images. By establishing the feature library of minutiae amd using fuzzy discriminate method, the few details characteristics were detected and then the situation of plant diseases and insect pests can be obtained. This method need only few detailed features to finish the detection, which avoides large pixel dependence. The experiment results show that the method can effectively detect most low quality images of plant diseases and insect pests, and achieve good results.

关 键 词：低质量图像 细节点库 模糊判别 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]