利用最近邻域分类的图像去噪算法  

Image Denoising Algorithm Using Nearest-Neighborhood Classification

作　　者：宫霄霖[1] 毛瑞全[1] 

机构地区：[1]天津大学电子信息工程学院,天津300072

出　　处：《天津大学学报》2011年第3期266-271,共6页Journal of Tianjin University(Science and Technology)

摘　　要：为了在图像去噪的同时较好地保护图像细节,利用最近邻域的算法将数据进行有效的分类,得到了不同的有意义的封闭邻域,从而对突出的边缘细节信息与非边缘细节信息进行有效地分割,较好地改善了方形邻域固定、模糊边缘细节信息的问题.且利用小波分析之后的系数特征,估计出一个最佳阈值并进行阈值去噪.实验表明,该算法可以得到更好的实验结果.To keep image details during image denoising,an effective algorithm was proposed to classify the data based on nearest-neighborhood method so as to get various close domains.In this way the salient edge and non-edge information can be divided effectively.This proposed method can improve the original method,which may blur the edge.An optimum threshold was estimated to de-noise the degraded image.The experiments show that the scheme can get better result than others.

关 键 词：图像去噪 脉冲耦合神经网络 图像分割 最近邻域 

分 类 号：TN391[电子电信—物理电子学]