摘要
为了提高芒果检测与分类的准确率和效率,综合运用计算机视觉技术和BP神经网络技术,实现对芒果损伤的检测与分类。首先,通过计算机视觉系统获取芒果图像,利用图像处理去除噪声、图像分割和图像增强等多种基本图像处理的方法,对芒果损伤图像进行处理;其次,对芒果图像进行了特征分析,提取9个特征参数,将这9个特征参数作为输入,建立BP神经网络模型,对芒果进行分类。试验结果表明,模型对芒果识别的准确率达85.5%。
In order to enhance the mango examination and the classification of accuracy and the efficiency, the system synthesis utilized computer vision technology, the BP artificial neural networks technology to realize the mango＇s damage detection and the classification. First, through the computer vision system gaining mango image, using the image processing elimination noise, the image segment, the image intensification and so on many kinds of basic image processing method carries on processing to the mango image. Secondly, we carried on the characteristic analysis to the mango image and picked up 9 characteristic parameters. Using these 9 characteristic parameters as the input, we established the BP neural network model, carried on the classification to mango＇s damage. The test result indicated the model to the mango recognition accurate rate has reached 85.5%.
出处
《农机化研究》
北大核心
2008年第10期57-60,共4页
Journal of Agricultural Mechanization Research
基金
广西研究生教育创新计划项目(2007105960811M24)
