中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
A New Method for Image Denoising with Nonlocal Means

文献类型:会议论文

作者Xuyao Zhang, Zhiyong Xu
出版日期2015
会议名称International Conference on Civil, Materials and Environmental Sciences (CMES 2015)
会议日期2015
页码99-102
中文摘要Denoising is an important problem in image processing, as it can influence the following processing step and decide the final visual effect. To denoise the image and preserve the details, this paper enhances the Nonlocal Means. When estimating the value of a disturbed pixel, we use the information of edges and textures as the weights of pixels involved in calculation, instead of simple Gaussian distance weights. The weights of edge and texture are obtained by direction grads and the coefficients of variation (CV). Finally, based on the results from different images, the enhanced algorithm is shown to have better performance than traditional NLM in SSIM, and has a better visual effect in subjective evaluations.
英文摘要Denoising is an important problem in image processing, as it can influence the following processing step and decide the final visual effect. To denoise the image and preserve the details, this paper enhances the Nonlocal Means. When estimating the value of a disturbed pixel, we use the information of edges and textures as the weights of pixels involved in calculation, instead of simple Gaussian distance weights. The weights of edge and texture are obtained by direction grads and the coefficients of variation (CV). Finally, based on the results from different images, the enhanced algorithm is shown to have better performance than traditional NLM in SSIM, and has a better visual effect in subjective evaluations.
语种英语
ISSN号2352-5401
源URL[http://ir.ioe.ac.cn/handle/181551/7705]  
专题光电技术研究所_光电探测与信号处理研究室(五室)
推荐引用方式
GB/T 7714
Xuyao Zhang, Zhiyong Xu. A New Method for Image Denoising with Nonlocal Means[C]. 见:International Conference on Civil, Materials and Environmental Sciences (CMES 2015). 2015.

入库方式: OAI收割

来源:光电技术研究所

浏览0
下载0
收藏0
其他版本

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。