基于结构张量的加权最小二乘纹理滤波

Weighted Least Squares Texture Filtering Based on Structure Tensor

针对图像纹理滤波需在保持结构的同时滤除多尺度纹理细节的问题,提出一种基于结构张量的加权最小二乘纹理滤波方法.该方法采用结构张量的各向异性和局部梯度提取图像中的结构信息,以此优化传统加权最小二乘滤波中每个像素点的平滑权重,这些权重决定了每个像素点的平滑程度.即在最小化全局优化框架中,结构处像素点的约束相对较小,因此赋予较小的权值;纹理区域内像素点的约束相对较大,因此赋予较大的权值.构造的多尺度保持结构边缘平滑算法框架,能将图像分解为分段平滑基础层和纹理细节层,得到自然图像的保持结构边缘平滑结果,以实现结构感知自适应纹理滤波的目的.与常用方法进行主客观对比的实验结果表明,该算法能有效保持结构边缘并平滑多尺度纹理,在结构相似度和视觉感知质量等方面均具有明显优势.

Aiming at the problem that image texture filtering need to preserve structures while filtering multi-scale texture details,we proposed a weighted least squares texture filtering based on structure tensor.The method used anisotropy of structure tensors and local gradient to extract the structure information in the image,so as to optimize the smoothing weights of each pixel in the traditional weighted least squares filtering,which determined the smoothness degree of each pixel.In the global optimization framework of minimization,the constraints of the pixels at the structures were relatively small,so smaller weights were assigned,the constraints of the pixels in the texture regions were relatively large,so larger weights were assigned.The constructed multi-scale edge-preserving smoothing algorithm framework could decompose the image into a segmented smooth base layer and texture detail layer to obtain the edge-preserving smoothing result of natural image,so as to realize the purpose of structure-aware adaptive texture filtering.The experimental results of subjective and objective comparison with the common methods show that the proposed algorithm can effectively maintain the structure edges while smoothing out multi-scale textures,and has obvious advantages in terms of structure similarity and visual perception quality.