项目名称: 基于Tetrolet变换的偏振遥感图像融合算法研究
项目编号: No.61272025
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 张德祥
作者单位: 安徽大学
项目金额: 61万元
中文摘要: 基于图像融合的偏振信息解析方法在目标跟踪和识别中可以显著提高目标在低对比度背景环境中被有效识别的能力。Tetrolet变换是采用二维经典Harr小波分解的多尺度多方向的各向异性变换,特殊的四格拼板分割的方法能在实现图像稀疏表达的同时保留偏振图像更多边缘和纹理,保持偏振特性的不变性。本项目拟开展基于Tetrolet变换的偏振遥感图像融合算法研究,包括以下四个方面:(1)研究Tetrolet高频系数的分布规律与图像细节的关系以明确对应偏振信息物理意义;(2)研究融合后多尺度高频系数的几何正则方向变化,实现四格拼版序号的重新定位以消除Gibbs现象;(3)研究多特征信息提取实现多尺度融合算法的选择(4)通过多特征融合提高低对比度偏振图像背景杂波的影响以提高视觉效果。拟通过以上研究,提出基于Tetrolet变换的偏振图像融合新方法,为实现高效、高精度和鲁棒的偏振遥感图像融合提供新的思路。
中文关键词: Tetrolet 变换;平稳Tetrolet 变换;偏振图像;图像融合;方块效应
英文摘要: For object tracking and recognition, the objects in the low contrast background are esier to be recognized by the image fusion-based polarization information analysis approaches. Tetrolet transform is a multiscale and multidirectional anisotropy transform using two-dimension classical Haar-type wavelet decomposition. For the tetrolet transform, a spare image representation can be provided by a special tetrominoes segmentation method. Meanwhile, more edge and texture information of the polarization image can be preserved, and the polarization character is still retained. In this project, we will carry out studies of Tetrolet transform-based polarization remote sensing image fusion algorithms. The main tasks of this project can be summarized as four aspects: (1) Analyze the relationship between the high frequency coefficient distribution and the image details in Tetrolet decomposition to determine the physical concept of the corresponding polarization information. (2) Investigate the geometrical regular direction changes of multi-scale high frequency coefficients, and relocate the tetrominoes serial number further to eliminate the Gibbs effect. (3) Investigate how to select the optimal multi-scale fusion algorithm through a multi-feature extraction strategy. (4) Investigate how to enhance the flooding wave of the
英文关键词: Tetrolet transform;Stationary Tetrolet transform;Polarization image;Image fusion;Block artifact