节点文献
一种高方差的高斯噪声滤波新方法
A Image Filtering Method for Gauss Noise of High Deviation Processing
【Author】 Shi Mei-hong, Yan Wei-wei, Zhang Jun-ying, Li Yong-gang(Computer Science Department, Xi’an University of Engineering Science & Technology, Xi’an 710048, ChinaKey Lab on Radar Signal Processing, Computer Institute, Xidian University, Xi’an 710071, ChinaComputer Department, Jiaxing University, Jiaxing 314001, China)
【机构】 西安工程科技学院计算机科学学院; 西安电子科技大学雷达信号处理重点实验室; 浙江嘉兴学院计算机系;
【摘要】 针对图像中高方差的高斯噪声的特点,提出了先定位、去除强噪声像素,后平滑弱噪声像素的滤波思想,即首先通过邻域像素值的变化特征,定位强噪声像素,并采用改进的自适应均值滤波方法滤除强噪声像素,然后利用改进的PCNN平滑弱噪声像素。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。
【Abstract】 A novel approach of image filtering is presented concerning Gauss noise with high deviation in this paper. This approach firstly locates strong noise pixels through estimating changes of pixel intensity in adjacent range and only removes these pixels using a modified adaptive mean filter, then smoothes weak noise pixels based on improved pulse-coupled neural networks. Experimental results show that proposed method works well with both preserving edge and smoothing range adaptively in an image, compared with other filtering methods
【Key words】 Gauss noise; Pulse-coupled neural networks (PCNN); Filtering; Edge protection; Adaptability;
- 【会议录名称】 第十二届全国图象图形学学术会议论文集
- 【会议名称】第十二届全国图象图形学学术会议
- 【会议时间】2005-10
- 【会议地点】中国北京
- 【分类号】TP391.41
- 【主办单位】中国图象图形学学会