【作者】 韩璠； 潘新； 郜晓晶； 闫伟红； 罗小玲； 刘霞；

【Author】 Han Fan;Pan Xin;Gao Xiao-jing;Yang Wei-hong;Luo Xiao-ling;Liu Xia;Inner Mongolia Agricultural University;Grassland Research Institute,Chinese Academy of Agricultural Sciences;

【机构】 内蒙古农业大学计算机与信息工程学院； 中国农业科学院草原研究所；

【摘要】 本文依据牧草叶片的纹理特征对18种牧草进行分类识别。针对牧草纹理具有分区的特点,采用分区LBP提取其纹理特征,借鉴蚁群算法作为分类器进行识别分类。实现了对牧草叶片图像的分类。将采集的样本经过标准预处理之后形成样本数据库。识别结果表明,最高识别率达到90%,同时实验结果证明:分区LBP能够对牧草叶片图像有效编码,蚁群算法可以提高牧草叶片识别的准确率,为实际应用提供了新的思路。更多还原

【Abstract】 We propose a novel algorithm for forage recognition of 18 species forages by extracting texture features.For every kind of forage texture has its partition character,the partition LBP and the ant colony algorithm are used to extract texture feature and to be classifier respectively.So the grass leaf images classification are realized.The collected samples were processed to form a sample database.The highest recognition rate achieves 90%.The experimental result shows that the partition-based LBP effectively encodes the grass images.The ant colony algorithm can also improve the accuracy of grass leaves recognition.更多还原