Blocking Artifacts Detection and Reduction in Block Based DCT Compressed Gray Scale Images

Abstract

The block based discrete cosine transform (BDCT) scheme takes in to account the correlation among pixels of the image. It usually divides the image into several 88 blocks, transforms each block from image pixels to DCT coefficients and then quantize the DCT coefficients. The blocks are coded separately and the correlation of pixels of two adjoining blocks is not taken into account in encoding. As a result, the block boundaries become visible when the decoded image is reconstructed. In other words, the two adjacent blocks which have a smooth change of luminous across the boundary, can result in a step shape in the decoded image if they fall into different quantization intervals. This kind of degradation is called blocking artifacts. In order to reduce blocking artifacts, blocking artifacts measurement is essential. In the proposed research work improves the approach presented by Singh et al. (2007) by adding the concept of filtering and mode detection. The blocking artifacts are removed without losing the image details by applying filters on the block boundaries,. The blocking artifacts are reduced to such an extent that the objective and subjective qualities are both improved. The computational complexity of proposed method is much lower than the method proposed by Singh et al. (2007).