关键词: 图像去噪, 混合噪声, 双树复小波变换, 自适应双边滤波器, 改进阈值

Abstract: For the current image denoising methods are mainly limited to the single noise, can not effectively solve the lack of the various mixed noise, an image denoising algorithm based on Dual-tree CWT and adaptive bilateral filtering is proposed. The algorithm uses the Dual-tree CWT to decompose the noisy image in multi-scales and multi-directions. The high-frequency coefficients of the sub-bands in each direction are quantized by the improved threshold, the low-frequency sub-band coefficients are filtered simultaneously by the adaptive bilateral filtering in each scale, and the reconstructed image is further removed. Experimental results show that the proposed method is superior to the existing algorithms, and can better protect edge detail information of the image. Through the quantitative comparison of the Peak Signal-to-Noise Ratio（PSNR） and the Root Mean Square Error（RMSE） of the objective evaluation index, PSNR can be improved by 5.333 2—6.527 8 dB, RMSE can be reduced up to 29.41%—46.03%, the running time is only 1.492 0 s, the overall noise reduction performance is more excellent.

Key words: image denoising, mixed noise, Dual-tree CWT, adaptive bilateral filtering, improved threshold