基于离散分数阶正交小波变换图像降噪新方法
发布时间：2018-12-11 09:24
【摘要】：分数阶小波变换是小波变换时间-频域的分析方法在时间-分数阶频率域的推广,在时间和分数阶频率域具有表征信号特征的能力.本文在离散分数阶正交小波变换(DFRWT)多分辨率分析(MRA)理论基础上,推导出DFRWT系数分解及重构新形式并作二维扩展.根据图像DFRWT子带系数能量随不同阶数p变化的特点,提出基于DFRWT阈值降噪新方法.该方法在保持子带低频能量为绝对大值条件下,适当提高子带高频能量值,更利于抑制图像噪声.实验结果表明,与传统小波阈值降噪方法相比,该方法主观质量得到了明显增强,提高了峰值信噪比.
[Abstract]:Fractional wavelet transform (FWT) is an extension of time-frequency domain of wavelet transform in time-fractional frequency domain. It has the ability to represent signal characteristics in time and fractional frequency domain. Based on the (MRA) theory of (DFRWT) Multiresolution Analysis of discrete Fractional orthogonal Wavelet transform, a new form of DFRWT coefficient decomposition and reconstruction is derived and extended in two dimensions. According to the characteristic that the energy of image DFRWT subband coefficient varies with different order p, a new method based on DFRWT threshold is proposed. Under the condition that the sub-band low frequency energy is kept as the absolute maximum, it is more advantageous to suppress the image noise by properly increasing the sub-band high frequency energy value. The experimental results show that compared with the traditional wavelet threshold denoising method, the subjective quality of the method is obviously enhanced and the peak signal-to-noise ratio (PSNR) is improved.
【作者单位】： 南京航空航天大学自动化学院;淮南师范学院电气信息工程学院;
【基金】：国家自然科学基金(No.60871009) 安徽省高校自然科学基金(No.KJ2011Z343)
【分类号】：TN911.72
[Abstract]:Fractional wavelet transform (FWT) is an extension of time-frequency domain of wavelet transform in time-fractional frequency domain. It has the ability to represent signal characteristics in time and fractional frequency domain. Based on the (MRA) theory of (DFRWT) Multiresolution Analysis of discrete Fractional orthogonal Wavelet transform, a new form of DFRWT coefficient decomposition and reconstruction is derived and extended in two dimensions. According to the characteristic that the energy of image DFRWT subband coefficient varies with different order p, a new method based on DFRWT threshold is proposed. Under the condition that the sub-band low frequency energy is kept as the absolute maximum, it is more advantageous to suppress the image noise by properly increasing the sub-band high frequency energy value. The experimental results show that compared with the traditional wavelet threshold denoising method, the subjective quality of the method is obviously enhanced and the peak signal-to-noise ratio (PSNR) is improved.
【作者单位】： 南京航空航天大学自动化学院;淮南师范学院电气信息工程学院;
【基金】：国家自然科学基金(No.60871009) 安徽省高校自然科学基金(No.KJ2011Z343)
【分类号】：TN911.72
【参考文献】
相关期刊论文 前4条
1 孟祥意;陶然;王越;;分数阶傅里叶域两通道滤波器组[J];电子学报;2008年05期
2 孟祥意;陶然;王越;;分数阶傅里叶域滤波器组的一般化设计方法[J];电子学报;2009年09期
3 陈U，
本文编号：2372303
本文链接：https://www.wllwen.com/kejilunwen/wltx/2372303.html
