多尺度几何分析阈值去噪的比较研究

A comparative study of Multiscale Geometric Analysis thresholding method

多尺度几何分析方法用于图像去噪已经成为图像处理方向的研究的一个热点。简要介绍和分析了几种常用于去噪的多尺度几何分析方法(曲波变换、轮廓波变换、剪切波变换),其中剪切波能提供了更稀疏的表示,图像实现最优逼近。实验结果也证明,剪切波变换阈值去噪在主观视觉上与峰值信噪比优于其他多尺度分析方法。

The Multiscale Geometric Analysis has become a hot spot in the current method of image denoising. After a brief introduction about some Multiscale Geometric Analysis methods in common use（Curvelet, Contourlet, Shearlet）, Shearlet transform provides more sparse decomposition than others, and can capture the best approximation of image.Experimental results show that Shearlet-based thresholding can obtain better performance in terms of both Peak Signal-to-Noise Ratio and subjective evaluations than other three Multiscale Geometric Analysis methods.