基于视觉传达技术的激光遥感图像匹配研究  

Research on laser remote sensing image matching based on visual communication technology

作　　者：胡海晓[1] 王艳贞[1] HU Haixiao;WANG Yanzhen(Shijiazhuang Tiedao University,Shijiazhuang 050000,China)

机构地区：[1]石家庄铁道大学,石家庄050000

出　　处：《激光杂志》2022年第4期98-102,共5页Laser Journal

基　　金：河北省重点研发计划项目(No.18215316D)。

摘　　要：为提升匹配精度,减少消耗的时空资源,研究基于视觉传达技术的激光遥感图像匹配方法。采用优化的Otsu自适应阈值算法对去噪图像实施边缘检测,提取共有特征,消除图像灰度差异性。采用傅里叶域检测的匹配方法得出匹配结果。实验结果表明:与传统图像匹配方法相比,新方法特征点分布均匀密集,匹配正确率可达98%,可实现图像去噪与高精度边缘定位。本方法在山区图像匹配中迭代次数为10,运行时间为1.268 s,在道路图像匹配中的迭代次数为7,运行时间为1.240 s,对比传统匹配方法,迭代次数和运行时间均得到降低,匹配效率得到提高。Aiming at the laser remote sensing image with translation, rotation and scale relationship, in order to achieve accurate image matching and reduce space-time resources consumed in matching process, a laser remote sensing image matching method based on visual communication technology is studied. According to the significant correlation and local sparsity of different bands in laser remote sensing image, a non local low rank dictionary learning model is constructed. The model is solved by a fast convergent alternating direction algorithm, and the image denoising is realized by using the redundant dictionary and sparse representation coefficient. The optimized Otsu adaptive threshold algorithm is used to detect the edge of the denoised image, remove the outliers, extract most of the common salient features in the image, and eliminate the gray difference of the features between the images. The Fourier domain detection method is used to obtain the translation parameters and correlation peak when there is rotation, translation or scale correlation between two images. The angle and translation parameters corresponding to the high peak are used as the matching results. The experimental results show that this method can effectively achieve image denoising and high-precision edge location, the same feature points in the matching results are evenly and densely distributed, and reduce the number of iterations and running time required in the matching process.

关 键 词：视觉传达 激光遥感 图像匹配 图像去噪 边缘检测 傅里叶域检测 

分 类 号：TN297[电子电信—物理电子学]