论文部分内容阅读
图像去噪是遥感图像复原的重要步骤。在去除图像噪声的同时希望尽可能多地保留图像的纹理细节信息。受较差的成像环境和图像数据远距离传输的影响,遥感图像中一般都含有较强的高斯-脉冲混合噪声,而在现有的图像去噪算法中,能够同时去除图像中的高斯-脉冲混合噪声的通用噪声滤波器很少。以非局部平均方法的滤波思想为基础,通过引入邻域相似度评价的概念和脉冲噪声探测器,提出了基于邻域特征匹配的通用噪声滤波器。实验结果表明:基于邻域特征匹配的通用噪声滤波器具备有很好地去除图像高斯-脉冲混合噪声的能力,在去除高斯-脉冲混合噪声的同时能够很好地保持图像的复杂纹理和精细细节,并且便于向DSP/FPGA多处理器平台上移植。
Image denoising is an important step in remote sensing image restoration. In the removal of image noise at the same time want to retain as much as possible texture details of the image information. Due to poor imaging environment and long-distance transmission of image data, remote sensing images generally contain strong Gaussian-pulse mixed noise. In the existing image denoising algorithm, the Gaussian-pulse There are few common noise filters that mix noise. Based on the idea of non-local average filtering method, a general noise filter based on neighborhood feature matching is proposed by introducing the concept of neighborhood similarity evaluation and impulse noise detector. The experimental results show that the general noise filter based on neighborhood feature matching has the ability to remove Gaussian-impulsive noise well and can remove the Gaussian-impulsive noise and keep the complex texture and detail of the image well , And easy to migrate to DSP / FPGA multiprocessor platforms.