Screen content video coding has become an emerging research topic with the spread of applications such as cloud gaming, screen/desktop virtualization, and mobile or external display interfacing. Screen content videos have different features compared to conventional camcorder-captured scenes. In this work, a novel low bit-depth representation-based motion estimation approach is proposed to exploit screen content specific features to improve coding efficiency. The proposed approach is based on an adaptive selection of Gray-coded bit-planes in order to generate low bit-depth representation of original screen content frames. The experimental results show that the motion estimation performance of the proposed approach is significantly better compared to the methods in the same category for screen content videos.