摘要
提出一种新的图像去噪方法,它是Besov空间的变分模型,在负实数次Sobolev空间上定义了数据项,用Besov半范数定义了正则项。并详细推导了变分模型在Besov空间的阈值求解公式,先做一个Contourlet变换域的小阈值收缩,然后再利用该模型去噪。去除噪声的同时也损失了部分边缘信息,把边缘分为四种情况,针对不同情况确定相应的边缘补偿方法。实验表明该模型具有良好的去噪效果。
A novel image denoising method is proposed in the paper.It is a variation model of Besov space.It defines both data items in negative real times Sobolev space and regularization items with Besov semi-norm.The paper derives in detail the threshold solve formula of variation model in Besov space.At first a small threshold shrinkage is made with Contourlet transform domain,then that model is applied for denoising.The denoising process causes four types of edge information loss,which are couterbalanced by different edge compensation methods respectively.Experiments show that the model performs well in denoising.
出处
《计算机应用与软件》
CSCD
2011年第9期9-11,37,共4页
Computer Applications and Software
基金
国家自然科学基金项目(60872157)
