SURE-LET Image Denoising with Directional LOTs

摘要：

This paper proposes to adopt hierarchical tree construction of directional lapped orthogonal transforms (Dir- LOTs) to image denoising. The DirLOTs are 2-D non-separable lapped orthogonal transforms with directional characteristics. The bases are allowed to be anisotropic with the fixed-criticallysubsampling, overlapping, orthogonal, symmetric, real-valued and compact-support property. As well, it is possible to introduce the trend vanishing moments (TVMs), which force wavelet filters to annihilate trend surface components. So far, the orthonormal wavelet image denoising techniques, such as the SURE-LET approach by Luisier et al., have shown a disadvantage in the restoration of diagonal textures and edges because of the separability of the adopted transforms. This work shows through some experimental results that the SURE-LET approach with DirLOTs overcomes the geometric problem.

作者: Shogo MURAMATSU Dandan HAN Hisakazu KIKUCHI

作者单位: Niigata University, 8050 2-no-cho, Ikarashi, Nishi-ku, 950-2181, Niigata, Japan

会议类型: 国际会议

会议名称: 2011亚太信号与信息处理协会年度峰会(APSIPAASC 2011)

会议地点: 西安

会议语种:英文

页码: 1-4

在线出版日期: 2011-10-18（万方平台首次上网日期，不代表论文的发表时间）