A Self-adaptive Non-local Means based on Segmentation and Wavelet

摘要：

　　Non-local means is a hotspot in image denoising, but this method can not estimate self-adaptive filtering parameters.Aiming at parameters chosen problem, we proposed a new NL-means which can adaptively estimate filter parameter.The basic idea of our method contains two points.First, we proposed a method which adaptively could estimate filter parameter c.Second, we segmented aimed-image to different texture-areas, and estimated parameter c corresponding different texture-areas.When filtering, aimed-area can get more contribution values from same texture-type areas.A series of tests show that our method can get more accurate denoising results, meanwhile obtain higher PSNR.

关键词： Non-local means wavelet adaptive filter parameter mean-shift segmentation

作者: Wenhui Li Bo Fu Ying Wang Huiying Li Hongyin Ni

作者单位: College of Computer Science and Technology, Jilin University, 130012, Changchun City, China

会议类型: 国内会议

会议名称: 第六届全国数字娱乐与艺术大会

会议地点: 长春

会议语种:英文

页码: 15-20

在线出版日期: 2011-09-25（万方平台首次上网日期，不代表论文的发表时间）