基于颜色和形状特征的棉花害螨图像分割方法

IMAGE SEGMENTATION METHOD FOR COTTON MITE DISEASE IMAGE BASED ON COLOUR AND SHAPE FEATURES

棉花害螨病是一种常见的棉花病害,以棉花害螨病为研究对象,提出一种在复杂背景条件下基于颜色特征和面积阈值的棉花害螨病斑的图像分割方法。算法主要分三步完成:类病斑(害螨病斑和茎杆)和绿色植物的分离、基于灰度直方图的单阈值分割、采用面积阈值法去除茎杆。实验结果表明,该算法能有效地提取出棉花害螨病斑,准确率可达94.79%。

Cotton mite disease is a common disease. Taking the cotton mite disease as the research subject, we propose an image segmentation method for cotton mite disease spots which is based on colour features and area thresholding under the condition of complex background. The algorithm has three stages： separation of the analogous disease spots （with mite disease spots and stem） from green plants, single thresholding segmentation based on gray histogram, and stems removal by using area thresholding. Experiment results show that this algorithm is effective in extracting the cotton mite disease spots, the correctness rate of the algorithm can reach 94.79%.