Image Denoising with Contourlet Transform Based on PCA

This paper adopted multiscale geometry method, distilled the principal component from the image after Contourlet transform, lowered the dimension of the high frequency subdomains, eliminated the noise by minimum variance cost function. The entire arithmetic without estimate noise, compared to Contourlet hard threshod denoising and wavelet hard threshod denoising, PSNR increased 1 dB, the denoising effect is better than other methods in the experiment.