论文部分内容阅读
本文提出了一种抑制合成孔径雷达图像乘性相干斑噪声的各向异性扩散滤波新方法.该方法将受自适应耦合函数控制的平均曲率运动嵌入到传统相干斑抑制的各向异性扩散方程中,形成了一种可有效抑制边缘区域相干斑与同质区块效应现象的各向异性扩散新方程,同时在新建的扩散方程中,引入了由改进Frost滤波与局部方向比率联合构建的一种带方向约束的新扩散函数,进一步削弱了块效应现象且明显改善了抑斑图像的边缘抖动扭曲问题.实验结果表明该方法在有效保护图像边缘的同时,能充分平滑同质区与边缘区域的相干斑,明显削弱块效应现象,有效改善抑斑图像边缘抖动扭曲问题,而抑斑图像无论视觉效果还是参数指标均比多种传统抑斑方法更具优势.
In this paper, a new method of anisotropic diffusion filtering to suppress the multiplicative speckle noise of synthetic aperture radar images is proposed. This method embeds the mean curvature motion controlled by the adaptive coupling function into the traditional anisotropic diffusion equation with speckle reduction , A new anisotropic diffusion equation that can effectively suppress the phenomenon of coherent patches and homogenous patches in the edge region is formed. In the newly established diffusion equation, a new improved diffusion equation is proposed by combining the improved Frost filter with the local directional ratio The new diffusion function with directional constraint further weakens the block effect and significantly improves the edge jitter distortion of the smudged image.The experimental results show that the proposed method can effectively smooth the edge of the homogeneity region and the edge region while effectively protecting the edge of the image The speckle images obviously weaken the block effect and effectively improve the edge jitter distortion of the speckle images. However, speckle images have more advantages than many traditional speckle reduction methods, both in visual effect and parameter index.