关键词: 图像处理, 去噪, 非下采样Contourlet变换, Context模型

Abstract: To overcome the Contourlet transform non translation invariance and spectrum aliasing defects, this paper presents a method based on nonsubsampled Contourlet transform for medical CT image denoising method. The noisy CT images are conducted by nonsubsampled Contourlet transform. Transform coefficients are obtained from different scales and different directions. Using Context model, subband of each scale and each direction is graded. Different classification uses the corresponding threshold denoising. Experiments show that this method is suitable to processing the medical CT image which contains more Gaussian noise. Compared with other methods, the PSNR value is improved, the image details are better retained, and CT image quality is improved.

Key words: image processing, denoising, nonsubsampled Contourlet transform, Context model