摘要
传统遥感图像阴影消除方法的图像噪声去除不彻底,导致阴影消除效果不理想。为此提出基于NSCT算法及Retinex理论的遥感图像阴影消除方法。划分阴影类型,从频率、亮度及颜色方面着重分析其属性。结合阴影属性,在两个不同色彩空间中对阴影检测。研究非下采样波轮廓变换分解过程,分别调整低频与高频系数,确定最终噪声阈值,利用NSCT算法完成图像去噪。利用自适应单尺度Retinex方法,经过参数设定、图像均衡化等操作,并在对数域中,将原始图像和滤波后的图像相减,获取阴影消除后的图像。仿真结果表明,上述方法阴影消除效果好,可获取更高质量的遥感图像。
In order to completely eliminate image noise,a shadow elimination method of remote sensing images based on the NSCT algorithm and Retinex theory is proposed.Shadow types were divided by their attributes(color,brightness,frequency).According to the shadow attribute,shadows were detected in two different color spaces.The decomposition process of non-down sampled wave contour transform was studied to adjust the low frequency and high-frequency coefficients respectively,obtaining the final noise threshold.The NSCT algorithm was utilized to eliminate image noise.The adaptive single-scale Retinex method was used.The parameters were set and the image was equalized.In the log domain,the original image was subtracted from the filtered image to obtain the shadow-eliminated image.The simulation results show that this method has an excellent shadow elimination effect.
作者
刘洋
张国军
LIU Yang;ZHANG Guo-jun(School of Microelectronics,University of Chinese Academy of Sciences,Beijing 100000,China)
出处
《计算机仿真》
北大核心
2022年第5期465-469,共5页
Computer Simulation
关键词
非下采样波轮廓变换
遥感图像
阴影消除
低通滤波
去噪处理
Non-subsampled wave contour transform
Remote sensing image
Shadow elimination
Low pass filtering
Denoising processing
