An efficient screen content video coding method to adaptively turn on/off filters used for intra-prediction is proposed. A block is classified into two categories based on a statistical property of a block and a transform coding mode in the compressed domain. Specifically, the Mahalanobis distance (MD) between the estimated probability density distribution and prediction samples in the left and top boundaries of the block to be coded and its transform coding mode is used. The set of the filters are disabled depending on the condition. Experimental results demonstrated that the proposed method outperforms the state-of-the-art video coding standard, i.e. the high-efficiency video coding/range extension (HEVC/RExt), in terms of rate-distortion (R-D) performance.