多尺度自适应阈值局部三值模式编码算法  

Coding algorithm of multi-scaled adaptive threshold local ternary pattern

作　　者：张雨 王强[1,2] 李柏林 高攀[1] Zhang Yu;Wang Qiang;Li Bailin;Gao Pan(School of Mechanical Engineering,Southwest Jiaotong University,Chengdu 610031,China;School of Mechanical Engineering,Chengdu Institute of Technology,Chengdu 611730,China)

机构地区：[1]西南交通大学机械工程学院,成都610031 [2]成都工业学院机械工程学院,成都611730

出　　处：《计算机应用研究》2020年第3期936-939,共4页Application Research of Computers

基　　金：四川省科技计划项目(2018GZ0361);四川省教育厅项目(16ZB0330)。

摘　　要：针对局部二值模式(local binary pattern,LBP)描述信息单一以及对噪声敏感的问题,提出一种多尺度自适应阈值局部三值模式(multi-scale adaptive local ternary pattern,MSALTP)编码算法。MSALTP首先将原始图像放大;其次将图像平均划分成几个区域,并计算像素的均值;然后计算每个区域中心像素与均值的偏差;最后提取ALTP特征,将结果统计特征直方图实现图像分类。实验表明提出的算法识别率比目前较好的抗噪声算法在不同的噪声下识别率有较大提高。Aiming at the single description and the noise sensitive problems of the LBP,this paper proposed a MSALTP algorithm.The algorithm first enlarged the original images.Secondly,it divided the images into several regions equally and calculated the mean value of the pixels.Then it calculated the deviation between the center and the mean pixel value of each region.Finally,it extracted the ALTP features and used the resulting statistical features histograms to classify the images.Experiments show that the proposed algorithm recognition rates are higher than the current anti-noise algorithms under different noise.

关 键 词：局部三值模式 自适应阈值 多尺度 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]