In this paper, to overcome the shortcomings of the state-of-the-art compression scheme for approximate multiplier, an optimized universal compression strategy for the compressor-based approximate multiplier has been proposed. When we apply the new compression scheme to the state-of-the-art compressors, accuracy of the approximate multiplier is largely increased with less hardware cost. Experiments on 8-bit and 12-bit approximate multipliers indicate that compared to the state-of-the-art, the new compression scheme can largely increase the precision by 19.6% and 18.1% respectively. When applied to image processing, the proposed multiplier also shows a better performance.