摘要
综合运用图像处理和模糊识别技术,以黄瓜病害为研究对象,进行黄瓜霜霉病自动识别的试验研究。在自然光条件下拍摄黄瓜叶片图像作为实验数据,为减少干扰因素对病害特征的不利影响,对原始图像做预处理,并把病斑分离出来;在农业植保专家的指导下,分析了黄瓜霜霉病病害的典型特征,从病斑形状、纹理和颜色三方面提取了16个特征参数;对黄瓜霜霉病叶片图像进行有监督的样本训练,得到黄瓜霜霉病害的标准特征模式,再对待测样本进行模糊聚类测试,平均识别准确率为95.28%。试验结果表明,该方法对于黄瓜霜霉病的识别效果较好。
Made scientific use of pesticides to enhance the quality of agricultural products,timely identify crop diseases,this paper taken cucumber diseases as the object,used two methods,image processing and fuzzy recognition,to study on the automatic identification of cucumber Downy mildew.Leaves images were taken under natural light as the experimental data.The original images were pre-processed and the lesions were extracted.Under the guidance of experts,the typical characteristics of the disease were analyzed,and 16 feature parameters were extracted from lesion shape,texture and color.For supervising training samples,a standard feature was got,and then test sample was treated for fuzzy clustering,the average accuracy rate of recognition was 95.28%.The results showed that the method was better.
出处
《中国农机化学报》
北大核心
2013年第2期212-216,共5页
Journal of Chinese Agricultural Mechanization
基金
吉林省世行贷款农产品质量安全项目(2011-Z20)
吉林农业大学青年启动基金(201126)
关键词
图像处理
模糊聚类
黄瓜霜霉病
image processing
fuzzy cluster
cucumber downy mildew
