摘要
对果林中的果害虫图像特征的分类识别,能够针对性的解决果林虫害问题。对果害虫图像特征的分类识别,需要计算出各个颜色特征分布标准差,提取图像的纹理特征,完成图像特征的分类识别。传统方法先计算出图像特征的显著值,对测试样本图像的特征向量进行类关联重构,但忽略了对图像纹理特征的提取,导致特征分类识别精度偏低。提出基于改进局部二进制模式的果林中果害虫图像特征高效分类识别方法。利用机器视觉方法提取害虫图像的RGB颜色空间和HSV颜色空间的特征,计算出各个颜色特征分布的均值和标准差,提取图像的纹理特征,反映害虫图像灰度分布均匀程度和纹理粗细程度,依据害虫图像特征类别抽取本质维数,实现果林中果害虫图像特征高效分类识别。实验结果证明,所提方法识别精度较高,为果林病虫草害的诊断提供了有利的依据。
An efficient method of classification and recognition of image characteristics of fruit pest in fruit - bear- ing forest is put forward based on improved local binary pattern. Machine vision method is used to extract the feature of RGB color space and HSV color space in pest images, and the mean value and standard deviation of distribution of various color characteristics are calculated to extract texture feature of image, which reflect the evenness degree of greyscale distribution and texture coarseness of pest image. According to classification of pest image feature, the in- trinsic dimension is extracted. Thus, the efficient classification and recognition of image characteristics of fruit pest in fruit - bearing forest is realized. Simulation results show that the proposed method has high accuracy of recognition, which provides the favorable basis for diagnosing pests in fruit - bearing forest.
出处
《计算机仿真》
北大核心
2018年第2期425-428,共4页
Computer Simulation
基金
自治区科技支撑项目(201531114)
关键词
林果害虫
图像特征
分类识别
Forest insect pest
Image features
Classification recognition
