From: Optimal graph edge weights driven nlms with multi-layer residual compensation
noise | NLM | BM3D | OGLR | ADNet | Our Method |
---|---|---|---|---|---|
\({\sigma }=10\) | 41.96dB 0.9797 | 45.20dB 0.9887 | 44.40dB 0.9859 | 43.63dB 0.9864 | 43.36dB 0.9779 |
\({\sigma }=20\) | 37.42dB 0.9742 | 40.88dB 0.9749 | 40.47dB 0.9705 | 40.19dB 0.9746 | 39.46dB 0.9601 |
\({\sigma }=30\) | 34.53dB 0.8983 | 38.13dB 0.9581 | 38.02dB 0.9522 | 37.80dB 0.9598 | 37.67dB 0.9543 |
\({\sigma }=40\) | 32.53dB 0.8463 | 36.08dB 0.9389 | 36.36dB 0.9368 | 36.17dB 0.9487 | 36.62dB 0.9581 |
\({\sigma }=50\) | 31.11dB 0.7915 | 35.16dB 0.9368 | 34.40dB 0.9046 | 34.84dB 0.9379 | 34.94dB 0.9445 |