基于总体变差模型的数字滤波器设计及其性能研究

Designing Digital Filter Based on Total Variation Model and The Research on its Performance

根据经典总体变差恢复模型,设计了一种新型的应用于图像去噪的数字TV滤波器的一种算法。该新型滤波器是绝对稳定的低通滤波器,在抑制噪声的同时能够很好的保留图像的细节信息。与一般的统计滤波器不同,该滤波器充分利用了局部数据的梯度信息精确计算滤波器系数,滤波过程是图像自适应的。文中直接给出了新型滤波器固定的模板形式,使滤波过程简单易行。最后将新型滤波器与中值滤波器、均值滤波器、高斯滤波器等进行性能比较,效果更好。

Based on the classical total variation restoration model, we design a new nonlinear algorithm of digital TV filter for image de-noising. This new filter is a low-pass filter with absolutely stability, capable of suppressing noise while preserving edges. It differs from the statistical filters, the filter coefficients can be calculated precisely by making full use of the gradient information in the local data, so the filter process is image adaptive. The filter is directly given in the form of template so that it is easy to implement. Comparing with other filters such as median filter, mean filter and Gaussian filter, our filter has better performance.