Abstract: It is the basis for the prevention and control of the crop diseases to detect accurately and rapidly the crop diseases according to the crop leaf symptoms. For accurate detection of the crop leaf diseases, a crop spot detection method is proposed based on Window Threshold Center-Symmetric Local Binary Pattern（WTCSLBP）. Firstly, the disease leaf features are obtained and the threshold to judge the spot is estimated by Adaptive Local Binary Pattern（ALBP）. Then the leaf image to be tested is divided into the same size detection window from which the ALBP features are also extracted. The ALBP features are compared to the threshold to find the disease spot. The experimental results on the database of three apple disease images show that the proposed method can effectively detect the crop disease spot distribution characteristics. Compared with Center-Symmetric Local Binary Pattern（CS-LBP） and WTCSLBP, this proposed method has the less number of dimension characteristics and higher correct recognition rate.

Key words: Local Binary Patterns（LBP）, Window Threshold Center-Symmetric Local Binary Pattern（WTCSLBP）, crop disease leaf, leaf spot detection

摘要： 根据作物叶片症状准确、快速检测作物病害是防治和控制作物病害的基础。为准确检测作物叶部病害，在窗阈值中心对称局部二值模式（WTCSLBP）的基础上，提出了一种作物病斑检测方法。首先利用自适应局部二值模式获取正常叶片图像特征并确定病斑判断阈值，然后将待检测叶片图像分割为大小相同的检测窗，并提取同样特征与阈值进行比较，以判断该检测窗是否有病斑。在三种苹果病害叶片图像数据库上的实验结果表明，该方法能够有效检测作物病斑分布特性。与中心对称LBP（CS-LBP）和WTCSLBP相比，该方法具有更少的特征维数和更高的正确识别率。

关键词: 局部二值模式（LBP）, 窗阈值中心对称LBP（WTCSLBP）, 作物病害叶片, 叶片病斑检测