基于小波多阈值和子带增强的图像去噪

Image Denoising Based on Wavelet Multi-thresholding and Subband Enhancement

为了在有效降低噪声的同时,尽量保留图像的边缘特征,提出了一种基于小波多阈值和子带增强的图像去噪方法.该方法对最小尺度小波系数采取软阈值方式,将其他小波系数再分解为近似子带和细节子带,依据误差度增强近似子带像素块,同时引入增强因子调节增强幅度;利用局部方差和混合阈值函数对各子带进行阈值处理,保证了图像达到较好的去噪效果.实验表明,与传统阈值方法相比,该方法不仅提高了去噪图像的峰值信噪比,而且较好地保留图像边缘特征,优于常规的阈值方法.

To maintain more edge features in the process of reducing image-noise effectively.A wavelet multi-thresholding for image de-noise associating with subband enhancement was proposed.The soft threshold operator removes the wavelet coefficients on a minimum scale.The other wavelet coefficients are divided into approximate subbands and detail subbands,then the pixel blocks of approximate subbands can be enhanced based on the error value;at the same time,the enhanced amplitude is well regulated by adding the plus factor.The image denoising effect is great by using local variance and hybrid threshold function for those subbands.The experimental results show that the proposed denoising method can increase the peak signal noise to ratio（PSNR） and maintain as many as possible the important edge features.Thus it has better performance than commonly used threshold method.