基于FSVM的二维经验模态分解域图像去噪

Image Denoising Using FSVM Classification in BEMD Domain

二维经验模态分解(Bidimensional Empirical Mode Decomposition,BEMD)是一种优秀的多尺度几何分析工具,能对非线性非平稳信号进行有效的分析。基于BEMD变换,提出了一种使用模糊支持向量机(Fuzzy Support Vector Machine,FSVM)的图像去噪算法。首先,应用BEMD变换把含噪图像分解成不同频率的子带;其次,BEMD系数通过FSVM训练被分成两类(无噪系数和噪声系数);最后,应用自适应阈值对含噪系数进行去噪。仿真实验结果表明,其算法不仅拥有较强的抑制噪声能力,而且具有较好的边缘保护能力。

BEMD(Bidimensional Empirical Mode Decomposition) is a kind of excellent multi-resolution analysis tool, and it can analyze nonlinear or non-stationary signal effectively. Based on BEMD transformation, an image denoising using fuzzy support vector machine(FSVM) is proposed. First, the noisy image is decomposed into different subbands of frequency using the BEMD transformation. Secondly, the BEMD coefficients are divided into two classes(edge-related coefficients and noise-related ones) by FSVM training model. Finally, the coefficients containing noise are denoised by using adaptive threshold value. Extensive experimental results demonstrate that the method can obtain better performances in terms of both subjective and objective evaluations than other current denoising techniques. Especially, the proposed method can preserve edges very well while removing noise.