Table 4 Rate savings of PE_LTW vs. Kakadu, SPIHT, and Kakadu with perceptual weights Kakadu_csf
From: Enhancing LTW image encoder with perceptual coding and GPU-optimized 2D-DWT transform
P_ELTW | vs. Kakadu | vs. SPIHT | vs. Kakadu_csf |
|---|---|---|---|
images | (% rate saved, mean) | (% rate saved, mean) | (% rate saved, mean) |
512×512 | |||
Lena | 13.87 | 16.83 | 5.23 |
Barbara | 11.39 | 17.44 | −2.61 |
Goldhill | 7.76 | 13.07 | 0.09 |
Boat | 8.58 | 12.02 | 0.47 |
Mandrill | 19.13 | 22.01 | 3.08 |
Balloon | 10.45 | 10.75 | 2.16 |
Horse | 14.96 | 14.91 | 3.74 |
Zelda | 17.22 | 20.43 | 8.46 |
Mean 512×512 | 12.92 | 15.93 | 2.58 |
2,048×2,560 | |||
Cafe | 9.63 | 12.34 | 1.43 |
Bike | 9.24 | 15.57 | −0.80 |
Woman | 5.21 | 11.46 | 3.75 |
Mean 2,048×2,560 | 8.03 | 13.12 | 1.46 |