基于改进Retinex算法的无人机影像匀光匀色应用  

Application of UAV image homogenization based on improved Retinex algorithm

作　　者：郑加富 巫春涛 ZHENG Jiafu;WU Chuntao(Reconnaissance and Design Institute of Qiantang River Administration in Zhejiang Province,Hangzhou,Zhejiang 310016,China)

机构地区：[1]浙江省钱塘江管理局勘测设计院,浙江杭州310016

出　　处：《测绘技术装备》2024年第2期62-67,共6页Geomatics Technology and Equipment

摘　　要：传统的单尺度Retinex匀光匀色算法使用的是高斯低通滤波器。针对高斯滤波函数作用在无人机影像频谱时,高频信息被截断,进而丢失部分影像细节,影响无人机影像匀光匀色效果的问题,本文提出了一种基于巴特沃斯低通滤波器的Retinex匀光匀色算法,使用巴特沃斯低通滤波器实现光照分量图的提取,保护了影像像素和细节,提高了无人机影像匀光匀色的效果。为了验证该算法的可靠性,本文还利用1景无人机影像进行试验,并与4种匀光匀色方法进行对比分析。结果表明,该算法在目视效果和定量评价指标上均呈现明显的优势,既能提高匀光匀色效果,又能保持影像的纹理和细节信息。The traditional single-scale Retinex homogenization algorithm uses Gaussi Low-pass Filter.When the Gaussian filter function is applied to the spectrum of the UAV image,the high-frequency information will be truncated,some image details will be lost,which will affect the light and color homogenization effect.So a Retinex homogenization algorithm based on Butterworth Low-pass Filter is proposed in this paper.Butterworth Low-pass Filter is used to extract illumination component map,which can protect image pixels and details and improve the effect of homogenization of light and color.The test was carried out using one scene UAV images to verify the reliability of the algorithm,and it is compared with other four methods of light and color homogenization.It is concluded that the algorithm proposed in this paper has obvious advantages in visual effect and quantitative evaluation index.The texture and details of the image can be maintained while the effect of light and color homogenization is improved.

关 键 词：无人机影像 匀光匀色 RETINEX 巴特沃斯低通滤波 

分 类 号：P231[天文地球—摄影测量与遥感]