【摘要】 基于决策的非对称裁剪中值滤波方法(MDBUTMF)能有效复原被高密度椒盐噪声污染的彩色图像,然而该方法采用固定大小的滤波窗口并利用均值替代中心像素,因此导致算法鲁棒性较低,对部分图像滤波失效。针对该问题,提出了一种基于自适应窗口的裁剪中值滤波方法,通过增加对单色区域的判断,有效解决了已有算法对单色区域滤波失效的问题,使得新方法具有较高的鲁棒性和实用性;并采用自适应窗口解决了MDBUTMF采用单一3×3窗口易导致滤波效果差的问题。实验数据表明,与经典的多种矢量以及标量的中值滤波方法相比,提出的裁剪中值滤波方法不仅具有较高的PSNR,而且具有较低的MAE和NCD,在抑制椒盐噪声的同时能有效保护图像的色调和细节。更多还原

【Abstract】 The decision-based unsymmetric trimmed median filter,which is efficient in restoring images contaminated by high-level pepper-salt noise,is however only effective to certain type of images due to its size-fixed filter window and the replacement of centre pixel using mean value of pixels in the window.Aiming at those drawbacks,this paper proposed an adaptive decision-based unsymmetric trimmed median filter.An estimation of homochromatic area is presented to solve the filter problem of original algorithm in homochromatic area.The noise restrain performance of MDBUTMF is improved by adopting adaptive size of filter window.Therefore,the proposed method is more robust and practical than original algorithm.It shows higher PSNR and also lower MAE and NCD in simulation comparing with some classic vector median filters and high-level noise restrain algorithms,restrains pepper-salt noise and retains image hue and details effectively in the meantime.更多还原