摘要
非下采样Contourlet变换(NSCT)不仅是一种多分辨率、局域的、多方向的图像变换方法,而且具有平移不变性。本文在其基础上对遥感图像去噪算法进行了研究,基于贝叶斯萎缩阈值法提出了一种自适应阈值选取方法,利用模糊集理论构造了一种自适应模糊阈值函数。分别在原始图像中添加了不同标准差的高斯白噪声和椒盐噪声,采用不同的去噪算法作为对比进行了实验。结果表明:该方法不仅能有效提高峰值信噪比,而且能够保持遥感图像丰富的纹理信息。
The Non-subsampled Contourlet Transform（NSCT） is not only an efficient directional multi-resolution image representation method,it is also fully shift-invariant.Based on NSCT.Based on the studies about the denoising algorithm of remote sensing images, an adaptive threshold denoising method was pro-posed by incorporating the Non-subsampled Contourlet Transform.The fuzzy theory was introduced in this method to construct an adaptive fuzzy threshold function.Gaussian white noise with different standard devi- ation value and pepper salt noise were added in the original image to get the noise images.The simulation results showed that the new approach could obtain higher peak-signal-to-noise ratio, meanwhile it could maintain the abundant edge information of the image.
出处
《遥感技术与应用》
CSCD
北大核心
2017年第3期435-442,共8页
Remote Sensing Technology and Application
基金
国家自然科学基金项目"城市边缘区地表组分温度反演模型的构建"(41571350)和"基于GPS/PS-DInSAR综合技术的城市大气水汽时空分布特征研究"(41301400)共同资助
关键词
NSCT
自适应阈值
去噪算法
纹理信息
NSCT
Adaptive threshold
Denoising algorithm
Texture information
