Ridgelet域中基于非参数自适应密度估计理论的图像去噪方法

Image De-noising Method Based on Nonparametric Adaptive Density Estimation in Ridgelet Domain

Ridgelet是一种新的信号分析方法,它适合于具有直线或超平面奇异性的二维信号的描述,目前,针对特定大小的离散图像,又提出了正交有限Ridgelet变换(FRIT)。该文在有限Ridgelet域中,结合Birge-Massart等提出的非参数自适应估计理论,提出一种新的二维图像去噪方法。实验证明,这种基于Ridgelet与Birge-Massart理论的图像去噪方法,与传统的Wavelet域去噪以及Donoho阈值去噪方法相比,去噪效果更为明显。

Ridgelet is a new signal analysis method; it is especially suitable for describing the 2-D signals which have linear or super-plane singularities. Recently, an orthonormal version off Ridgelet for discrete and finite-size images is presented, named Finite Ridgelet Transform （FRIT）. In this paper, a new image de-noising method is proposed by using the threshold method based on nonparametric adaptive estimation which is presented by Birge-Massart in Ridgelet domain. Experiments show that this de-noising method represents better characteristic than traditional de-noising method in wavelet domain and the de-noising method based on Donoho strategy.