Nonsubsampled contourlet-based anisotropic diffusion filtering for infrared image denoising

Chao Ni,Qi Li,Liangzheng Xia
2010-01-01
Abstract:According to the characteristics of infrared image, a method of nonsubsampled contourlet-based anisotropic diffusion filtering for infrared image denoising is proposed. First, nonsubsampled pyramid technique is used to decompose noisy image into lowpass subband and multi-scale bandpass subbands. Nonlinear anisotropic diffusion filtering is performed in all bandpass nonsubsampled pyramid layers to aim at preserving the edges and details when reducing noise. Then nonlinear diffusion filtered bandpass nonsubsampled pyramid layers are decomposed by nonsubsampled directional filter banks. Nonsubsampled contourlet coefficients are processed by thresholding for inter-scale correlations, and the denoised image is achieved. Experiments show that the method can effectively achieve better SNR(signal to noise ratio) and visual effect for edge preserving of denoised infrared images in comparison with other methods.
What problem does this paper attempt to address?