摘要
利用特征选择后粮仓害虫的10个形态学特征,在归一化分析的基础上,利用特征的均值和标准差来构造粮虫的经典物元和节域物元,提出了基于模糊分析定量确定特征权重系数的新方法.在计算待识别粮虫与各类粮虫的关联度的基础上,依据最大关联度准则对储粮害虫进行分类判别.并对粮仓中危害严重的9类粮虫进行了自动分类,识别率达到93%以上,结果表明,依据均值和标准差来构造粮虫的经典物元和节域物元可进一步提高识别的精度.
The ten morphological features of the stored-grain pests were normalized after selecting features. The standard and extensional matter-element matrixes were constructed based on the feature mean value and standard deviation. A quantitative method identifying the feature weight coefficients by fuzzy analysis was put forward. The correlative degrees between the stored-grain pests to be recognized and the nine species pests were calculated, such that the pests were classified according to the principle of the maximum correlative degree. The nine species of the stored-grain pests in grain-depot were automatically recognized by a classifier based on the extension decision theory, and the identification ratio was over 93%. The experiment showed that the recognition ratio can be improved by constructing standard and ex- tensional matter-element matrixes based on the feature mean value and standard deviation.
出处
《江苏大学学报（自然科学版）》
EI
CAS
北大核心
2008年第4期284-287,共4页
Journal of Jiangsu University：Natural Science Edition
基金
江苏大学现代农业装备与技术国家重点实验室培育点开放基金资助项目(NZ200707)
江苏大学博士生创新基金资助项目
关键词
储粮害虫
可拓理论
权重
模糊逻辑
图像识别
stored-grain pests
extension theory
weight
fuzzy logic
image recognition
