Blocky artifacts reduction by noise estimation in DCT coded images

A postprocessing algorithm is proposed to reduce the blocky artifacts of Block Discrete Cosine Transform (BDCT) coded images. To analyze blocky artifacts as noise components residing across two neighboring blocks, we use one-dimensional pixel vectors made of pixel rows or columns across the neighboring blocks. We model the blocky noise in each pixel vector as shape vector multiplied by boundary discontinuity. We estimate the shape vectors, which vary with local image activities, by MMSE(Minimum Mean Squared Error) approach. Then the proposed algorithm eliminates the noise components in each pixel vector according to its boundary discontinuity and local image activity. The experimental results show that the proposed algorithm is highly efficient in reducing the blocky artifacts in both subjective and objective viewpoints.