基于最大类间方差法的黄瓜病害叶片分割

摘要

Taking diseased cucumber leaves as research object,the background was removed by the histogram threshold segmentation method.The diseased parts of cucumber leaves were segmented by using Otsu method and edge detection method.Otsu method was improved and be used to segment red component of disease spots.The results showed that the profile of the disease spot segmented by edge detection method was incomplete,and the segment effect of Otsu method was better.A better segmentation effect of diseased cucumber leaves was obtained by using Otsu method,which laid the foundation for identifying disease in the future.%以黄瓜病害叶片为研究对象,采用直方图阈值分割法去除背景,分别采用最大类间方差法(Otsu)和边缘检测法来分割黄瓜叶片中的病害部位,对比这2种方法的分割效果.最后,对已有的最大类间方差法进行了改进,对病害叶片图像的红色分量进行了病斑分割.结果表明,边缘检测法分割出来的病斑部位轮廓具有不完整性,而最大类间方差法的分割效果较好.采用最大类间方差法对黄瓜病害叶片分割取得了较理想的效果,为后续病害识别奠定了基础.