In decoded digital video, the local perceptual compression artifact level depends on the global compression ratio and the local video content. We propose a new algorithm for block artifact reduction for which Machine Learning is used. In H.264 Video Compression Standard, an In-loop Deblocking filter is implemented to reduce the quantization noise around the block boundaries. But there would still remain artifacts with in blocks. These artifacts are known as Contour artifacts. These Contour artifacts further be reduced by our Proposed Machine Learning Artifact Reduction Method. The Proposed method is selectively applied to macro-blocks damaged by block artifacts. Its effectiveness will be clearly demonstrated for all the evaluation aspects.We estimate the level of block artifacts by measuring PSNR of the block artifacted Video. Our created reference metric is an accurate local estimator of the compression artifact level. We were able to copy the performance to two no-reference metrics, based on a weighted mixture of low-level features. Our new metrics enable a far superior performance of artifact reduction compared to relevant alternative proposals.