摘要
针对图像中高方差的强高斯噪声特点,提出了一种图像噪声联合滤波的新方法。算法将受强高斯噪声污染的图像分为强噪声点集和弱噪声点集两部分,首先通过邻域像素强度值的变化特征,定位强噪声像素点,并采用改进的自适应均值滤波方法滤除,然后基于简化的脉冲耦合神经网络(PCNN)平滑弱噪声点像素。经实验结果验证,与已有的其他滤波方法相比,该算法在较好地滤除噪声的同时,具有良好的图像边缘保护和自适应能力。
A new joint method for filtering image contaminated with strong Gaussian noises was presented. The pixels of an image were divided into two sets. Strong noisy pixels were located firstly through estimating changes of pixel intensity in a local region of a pixel and were removed using a modified adaptive mean filter. Weak noisy pixels were smoothed with a simplified pulse-coupled neural network (PCNN). Experimental results show that the proposed method works well with both preserving edge and smoothing range adaptively in an image, compared with some existing image filtering methods.
出处
《计算机应用》
CSCD
北大核心
2007年第7期1637-1640,共4页
journal of Computer Applications
基金
陕西省教育科研项目(05JC13)
陕西省科技厅国际合作资助项目(2004WK-06)
关键词
高斯噪声
脉冲耦合神经网络
滤波
保护边缘
自适应
Gaussian noise
Pulse-Couplod Neural Network (PCNN)
filtering
preserving edge
adaptability
作者简介
石美红(1956-),女,江苏仪征人,教授、主要研究方向:图像处理、模式识别、智能信息处理;meihong_shi@263.com
毛江辉(1983-),男,江苏丹阳人,硕士研究生,主要研究方向:图像处理、并行计算;
梁颖(1980-),女,陕西汉中人,硕士研究生,主要研究方向:计算机网络安全;
龙世忠(1983-),男,江西吉安人,硕士研究生,主要研究方向:图像处理、并行计算。