T. Goto, M. Ishikawa, and T. Kitamura (Japan)
Blocky noise, Discrete cosine transform, Wavelet transform, Noise reduction
JPEG and MPEG International Standards have been used as a compressed and stored format for still images and moving images, respectively. In these standards, there is a prob lem with blocky noise appearing in reconstructed images when a higher compression rate is set. This is because there are no correlations between blocks as a result of quan tizing low band signals. So in the traditional method, the reconstructed images with blocky noise are filtered with a low-pass filter. As a result, blocky noise can be decreased. However, by filtering the reconstructed image, the edge sig nal is also smoothed out, and the resulting image is blurry. In this study, we will propose a new method of re ducing blocky noise using inverse wavelet transform. In the proposed method, JPEG images, which are low resolu tion and include blocky noise, are transformed to wavelet transform coefficients. And a masking process using these coefficients is done. As a result of applying the proposed method to a decoder, blocky noise is decreased and the re sulting image is clear.
Important Links:
Go Back
