一种针对彩色图像随机值脉冲噪声的矢量滤波方法

A Vector Filtering Method Aimed at Color Image Random-valued Impulse Noise

为了去除彩色图像随机值脉冲噪声,提出了一种新的矢量滤波方法。该方法对图像的平滑区域和边缘区域的滤波工作分开进行,平滑区域滤波方法将窗口分成多个区域,然后基于矢量中值和平滑区域像素的特征检测出平滑区域的信号,边缘区域的滤波是在已知信号的基础上对非信号进行矢量中值滤波。仿真实验结果表明,该方法能够有效地去除彩色图像的随机值脉冲噪声,尤其当噪声密度较高时,去噪效果明显优于传统的矢量中值滤波。

To remove random-valued impulse noise of color image, this paper presented a new vector filter. This method remove flat-region and edges noise respectively, the filter window is divided into regions by flat- region filter, then flat-region filter detects flat-region signals based on vector median filter and characteristics of flat-region pixels, edge filter is vector median filter based on known signals. The simulation result show that, this method can remove random-valued impulse noise of color image efficiently, when the images are highly corrupted by impulse noise, this method gives superior performance compared with vector median filter.