摘要：

来自图像传感器的数字图像会受到各种噪声的干扰，其中主要包括加性噪声、乘性噪声和混合噪声。乘性噪声随信号幅度改变而改变，没有理想的去除方法。为此，运用基于总体最小二乘法的图像估计降噪方法，研究图像块尺寸选取对降噪性能的影响，分析成像系统中去马赛克环节影响噪声传播的内在规律，并通过比较实验给出总体最小二乘法降噪的性能优势。

关键词: 图像降噪, 去马赛克, 总体最小二乘法, 乘性噪声

Abstract:

Images from the digital image sensor are corrupted by all kinds of noise interference, including additive, multiplicative and mixed noise. In particular, the extent of multiplicative noise changes with signal amplitude, there is no better removing method, and Total Least Squares (TLS) estimation is used on image de-noising in this paper. The size of the image patches impacting more greatly on estimation is studied. In addition, de-mosaic impacting on noise propagation in image pipelines is analyzed, and the overall advantages of de-noising using TLS method are comparatively given.

Key words: image de-noising, e-mosaic, otal Least Squares(TLS), ultiplicative noise

中图分类号: 

• N945