基于融合的DWT-LTP-IGLCM的野生食用菌纹理特征分类方法研究  

Research on classification method of texture features of wild edible fungi based on fusion DWT-LTP-IGLCM

作　　者：张国霜 王娜 李卓识[2] 林楠[2] Zhang Guoshuang;Wang Na;Li Zhuoshi;Lin Nan(College of Biomedical Engineering,Jilin Medical University,Jilin,132000,China;College of Information Technology,Jilin Agricultural University,Changchun,130118,China)

机构地区：[1]吉林医药学院生物医学工程学院,吉林吉林132000 [2]吉林农业大学信息技术学院,长春市130118

出　　处：《中国农机化学报》2020年第8期108-115,共8页Journal of Chinese Agricultural Mechanization

基　　金：吉林省自然科学基金(20180101041JC);吉林省教育厅“十三五”科学技术项目(JJKH20180659KJ);吉林省高等教育学会高教科研重点项目(JGJX2019B16)。

摘　　要：为更好解决野生食用菌纹理特征提取方法的识别效率和准确性问题,针对相似野生食用菌褶,提出DWT-LTP-IGLCM的纹理特征提取融合方法,利用DWT有效保留图像的主要特征,通过LTP中心像素与相邻像素间的关系设定固定的阈值,不仅消除了噪声的干扰且具有更好的光照鲁棒性,结合旋转不变的灰度共生矩阵IGLCM的特征参数,得到了12个食用菌纹理特征值进行分析,基于支持向量机SVM方法对其特征进行有效的分类。试验结果表明:DWT-LTP-IGLCM的纹理特征提取,可以有效将相似食用菌进行分类识别,分类的正确率达到了100%,并且减少了食用菌外部特征分类的错分率,提高了食用菌分类识别效果。In order to better solve the problem of identification efficiency and accuracy of texture feature extraction method for wild edible fungi.For the similar wild-edible mushroom pleat,a texture feature extraction and fusion method of DWT-LTP-IGLCM was proposed.The main features of the image were effectively retained by DWT,and a fixed threshold was set through the relationship between LTP central pixel and adjacent pixel,which not only eliminated noise interference but also had better illumination robustness.Combining the feature parameters of IGLCM,a rotation-invariant gray scale symbiosis matrix,12 texture feature values of edible fungi were obtained for analysis,and their features were effectively classified based on SVM method.The experimental results showed that the texture feature extraction of DWT-LTP-IGLCM could effectively classify and identify similar edible fungi,and the classification accuracy was up to 100%.In addition,the misclassification rate of external characteristics of edible fungi was reduced,and the classification and recognition effect of edible fungi was improved.

关 键 词：纹理特征提取 分类识别 支持向量机 野生食用菌 

分 类 号：S58[农业科学—作物学] S646