【摘要】 特征提取和分类器的设计是田间害虫图像识别中的关键环节。针对害虫目标的二值化图像提取出面积、周长、复杂度等7个形态学特征,并进行归一化处理;建立了9种害虫的模板库及隶属度函数,并基于最小最大的原则进行模糊决策分析;对稻纵卷叶螟、棉铃虫等田间危害严重的9种害虫进行识别分类的识别率达86%以上。更多还原

【Abstract】 The feature extraction and the classification design are the key parts in the image reco-gnition for agricultural pests.Seven morphological features from the binary images of agricultural field pests,such as area,perimeter,and so on were extracted and normalized.The data templates of nine pieces of pests and the subject function were set and the fuzzy decision is analyzed based on the fuzzy minimums and maximums rule.The result shows that the fuzzy classification can be nine pieces of pests that harmed seriously,such as rice leaf folder,cotton bollworm,etc.The correct recognition accuracy is over 86%.更多还原