多尺度域非线性能量保持的图像去噪新方法

New image denoising method based on multi-scale decomposition and nonlinear energy preserve

提出了一种基于Contourlet多尺度分解域核主成分分析的图像去噪新方法。该方法首先对源图像进行Contourlet分解,在不同频段的子带图像中,利用核主成分分析方法进行能量保持,用重构图像来进行子带去噪,最后通过Contourlet逆变换得到去噪之后的图像。实验结果表明,该方法不仅有效地降低了图像噪声,且峰值信噪比也较高。

A new method based on Kernel Principal Component Analysis（KPCA） in contourlet multi-scale decomposition domain is proposed in order to solve image denoising problem.Firstly,contourlet transformation is applied on source image with this method.Secondly,image denoising is executed on different frequency sub-images with reconstructed images by KPCA method.Finally,denoised image is attained with inverse contourlet transform.The experimental results on images demonstrate that the proposed method not only decrease image noise effectively,but also improve PSNR.