中文摘要：

根据甘蔗苗期赤腐病和环斑病图像的特点,提出了一种甘蔗病害图像分割方法。首先利用颜色特征2G-R-B和2R-G-B提取出病斑和土壤等非绿色植物类。然后采用面积阈值分割法排除部分土壤等非绿色植物类连通区域。最后利用链码计算剩下的病斑和土壤等非绿色植物类连通区域的形状特征,根据区域的宽度、矩形度和圆度分离出病害病斑。实验结果表明,该算法能有效提取出赤腐病和环斑病病斑,对环斑病图像分割正确率达93%,对赤腐病图像分割正确率达95%。

英文摘要：

An image segmentation algorithm for the sugarcane diseases was proposed according to the characteristics of the sugarcane seedling red rot disease and ring spot disease images. Firstly, nongreen plants including diseased spots, soils etc. were extracted using the color features of 2G -R -B and 2R-G-B. Secondly, some connected regions of non-green plants including soils etc. were removed using the segmentation based on area threshold. Finally, the shape features for the remaining connected regions of non-green plants including diseased spots, soils etc. were calculated using chain code, and the diseased spots were extracted according to the width, rectangle degree and roundness of the connected regions. The experimental results showed that this algorithm could effectively extract red rot diseased spots and ring spot diseased spots, the accurate rate of image segmentation for ring spot diseases was 93 %, the accurate rate of image segmentation for red rot diseases was 95 %.