Retinex算法在无人机遥感影像匹配中的应用  

Application of Retinex Algorithm in Remote Sensing Image Matching of Unmanned Aerial Vehicle

作　　者：张莉 祁琼 孔姗 刁婷婷 ZHANG Li;QI Qiong;KONG Shan;DIAO Tingting(Hubei Spatial Planning Research Institute,Wuhan 430071,China)

机构地区：[1]湖北省空间规划研究院,湖北武汉430071

出　　处：《地理空间信息》2024年第10期30-33,共4页Geospatial Information

摘　　要：无人机遥感测量系统数据获取时由于外部环境的影响,获取的影像内部及影像间色彩差异大,影响影像特征点提取数量及均匀度,降低了影像定向参数解算的可靠性,为此将MRS(multi-scale retinex)算法与高斯滤波算法相结合,提出了MRSGF(multi-scale retinex with gaussian filters)方法对无人遥感影像进行增强。使用偏暗、偏亮、光照分布不均的无人机遥感影像对MRSGF进行验证,实验表明该算法提高了SIFT(scale-invariant feature transform)算法对无人机航空影像纹理多样性及色彩差异的适应性,提高了SIFT算法特征点的数量及分布的均匀度。Due to the influence of external environment,the color differences within and between acquired images are large,which affects the number and uniformity of image feature point extraction,reducing the reliability of image orientation parameter resolution.Therefore,combining the MRS(multi-scale Retinex)algorithm with the Gaussian Filter algorithm,we proposed a MRSGF(multi-scale Retinex with Gaussian Filters)method to enhance UAV remote sensing images.Then,we used dark,bright and unevenly illuminated UAV remote sensing images to validate the MRSGF algorithm.The experimental result shows that the algorithm improves the adaptability of SIFT(scale-invariant feature transform)algorithm to the texture diversity and color differences of UAV aerial images,and improves the number and distribution uniformity of SIFT algorithm feature points.

关 键 词：无人遥感 SIFT MSR MRSGF 

分 类 号：P231[天文地球—摄影测量与遥感]