摘要: |
提出了一种基于序列二次规划(SQP)优化阈值的非下采样Contourlet变换(NSCT)图像高斯白噪声去除方法。该方法利用广义交叉验证(GCV)准则作为优化指标,使用序列二次规划算法对NSCT域的去噪阈值进行优化,能够在噪声方差等图像先验知识未知的情况下得到最优去噪阈值。确定阈值后,采用非线性阈值函数对Contourlet系数进行处理。实验结果表明与其他Contourlet域去噪方法相比,该方法能有效去除图像的高斯白噪声,提高图像的峰值信噪比,并较好地保留图像的边缘信息。 |
关键词: 图像去噪 非下采样Contourlet变换 广义交叉验证 序列二次规划 |
DOI: |
|
基金项目:国家自然科学基金(61503009,61333011,61421063);中国航空科学基金(2016ZA51005);上海航天科技创新基金(SAST2016003) |
|
An Image Denoising Method Based on Nonsubsampled Contourlet Transform with SQP Optimization |
YANG Chen,DON GXi-wang,LI Qing-dong,REN Zhang |
(School of Automation Scicence and Electrical Engineering,Science and Technology on Aircraft Control Laboratory, Beihang University, Beijing 100191, China) |
Abstract: |
An image denoising method based on nonsubsampled contourlet transform (NSCT) with successive quadratic programming (SQP) optimization is proposed. This method can obtain the optimal threshold for each subband without the priori information of the noise variance using SQP optimization and generalized cross validation (GCV) criterion. After the threshold is determined, a nonlinear threshold function is applied to overcome the inadequate of soft threshold and hard threshold function. The experimental results show that the proposed method has a better performance than other contourlet-based image denoising methods and outperforms on both visual quality and peak signal-to-noise ratio (PSNR). |
Key words: Image denoising Nonsubsampled contourlet transform Generalized cross validation Successive quadratic programming |