论文部分内容阅读
针对万寿菊黑斑病难于防治的问题,采用基于主成分分析和BP神经网络的识别方法,对万寿菊黑斑病病原菌(Alternaria tagetica)无侵染力和有侵染力的孢子进行精确识别。首先利用图像处理技术对病原菌孢子显微图像进行分割,选取3个颜色特征(R、G和V)、5个形状特征(Hu不变矩中的H2、H3、H4、H5和H6),以及3个纹理特征(R、G、B3个分量的对比度)共11个特征用于病原菌孢子分类识别。为提高识别速度和精度,利用主成分分析法(PCA)对11个特征进行优化和筛选,采用基于L-M算法的BP神经网络对万寿菊黑斑病病原菌的孢子进行分类识别。试验结果表明,经主成分分析后得到的第一、第二主成分能够有效减少BP网络训练时间和提高识别准确率,平均识别准确率达到98%。该方法能够精准识别万寿菊黑斑病病菌有侵染力和无侵染力的孢子。
Aiming at the difficulty of preventing and curing the black spot disease of marigold, the non-invasive and infecting spores of Alternaria tagetica were identified accurately based on principal component analysis and BP neural network. Firstly, the microscopic images of pathogen spores were segmented by using image processing technology. Three color features (R, G and V), five shape features (H2, H3, H4, H5 and H6 in Hu moments) Three texture features (R, G, B3 component contrast) A total of 11 features for pathogen spore classification identification. In order to improve the speed and precision of identification, 11 features were optimized and screened by principal component analysis (PCA), and the spores of black spot pathogens were identified by BP neural network based on L-M algorithm. The experimental results show that the first and second principal components obtained after principal component analysis can effectively reduce the training time of BP network and improve the recognition accuracy, with an average recognition accuracy of 98%. This method is able to precisely identify the infective and non-infective spores of the marigold black spot pathogen.