A Fast DCT Block Smoothing Algorithm S

t accuracy. We have found a simplification of our previous methods that performs nearly as well as hose methods and that can be computed much more rapidly and locally. e o When errors are clearly visible, the blockiness of some artifacts distinguishes them from th riginal image content, suggesting a way of reducing these artifacts. Presumably, the human viewer k e identifies the artifacts by assuming the image does not have 8-pixel by 8-pixel block features. Bloc dge variance, the sum of squared differences between adjacent block edge pixels, is one simple s m measure of blockiness. A similar measure offset from the block edge provides an estimate of what thi easure would be if the image were not blocky. In previous work, we found that using a global 5, 6