摘要
偏振成像技术能够反映出传统光学成像所无法反映的物体特征,有效地增强物体特征的对比度。针对传统纹理特征提取方法加权平均光强图像4种空间信息下的灰度共生矩阵特征,造成图像纹理特征细节缺失,难以从灰度对比差异角度完整地提取图像纹理特征等问题,结合偏振信息,提出一种偏振灰度差异共生矩阵对偏振图像纹理特征进行提取,并以偏振灰度差异共生矩阵的特征为研究依据,分别推导出均质性、离散性、粗糙性、相异性4种特征参量作为图像的偏振纹理特征参量。旋转偏振相机前面的偏振片获取偏振光强图像并进行实验,结果表明,偏振光强图像纹理特征的均匀程度、离散程度、粗糙程度和相异程度在不同偏振方向上不会反映出较大差异。通过偏振灰度差异共生矩阵提取到的4种偏振纹理特征参量图像,随着偏振方向的变化,都反映出了显著差异。
Different objects or different parts of the same object have different polarization information due to varied roughness,materials,contours and viewing angles.In this paper,image analysis of polarized images is made to explore the close connection between the polarization information and the texture statistical properties for the purpose of texture characterization.In this paper,polarized light intensity images in four polarization directions are obtained by the rotation of the polarizer in front of the polarization camera.Then,the polarization quadrature difference operation is employed to eliminate the natural light component information,obtain the polarization difference images containing more polarization information,and conduct research on the texture statistical properties of the polarized images.With reference to the image texture feature analysis,the polarization information comes into use to optimize the traditional gray level co-occurrence matrix,and calculate the co-occurrence matrix of polarized images.Thus,the texture features of the polarized images is obtained and the directional anisotropy of the texture features of the polarized images is enhanced.Two major tasks are accomplished in this paper:first,the polarization gray level difference co-occurrence matrix is proposed.It effectively combines the polarization information and analyzes the contrast difference between the gray level of the pixel on the images and the gray level of neighboring pixels in the polarization direction,enriching polarization information and spatial information simultaneously,overcoming the limitations of the gray level co-occurrence matrix in the traditional texture feature analysis,which are manifested by failure to consider the gray contrast difference between pixels under certain spatial information and thus arising difficulties to thoroughly describe the texture characteristics of the images from the perspective of gray contrast difference;second,based on the features of the polarization gray level difference co-occurr
作者
刘云清
姜晓娇
段锦
莫苏新
张文学
LIU Yunqing;JIANG Xiaojiao;DUAN Jin;MO Suxin;ZHANG Wenxue(School of Electronic and Communication Engineering,Changchun University of Science and Technology,Changchun 130012,China)
出处
《重庆理工大学学报（自然科学）》
北大核心
2023年第10期117-126,共10页
Journal of Chongqing University of Technology：Natural Science
基金
国家重点研发计划项目(2022YFC2203901)。
关键词
偏振信息
纹理特征
偏振灰度差异共生矩阵
偏振纹理特征参量
polarization information
texture feature
polarization gray level difference co-occurrence matrix
polarization texture feature parameter
