Table 1 Experimental data set and performance comparison of original and improved CUDA-kFOE
From: An improved parallel fuzzy connected image segmentation method based on CUDA
Dataset | Small | Medium | Large |
|---|---|---|---|
Seed position | (166, 224, 88) | (189, 245, 175) | (220, 217, 497) |
Scene domain | 512Â *Â 512Â *Â 131 | 512Â *Â 512Â *Â 261 | 512Â *Â 512Â *Â 576 |
Voxel size (mm3) | 0.69Â *Â 0.69Â *Â 1.0 | 0.70Â *Â 0.70Â *Â 1.0 | 0.87Â *Â 0.87Â *Â 0.8 |
CPU time (s) | 386 | 783 | 1157 |
Origin GPU version (s) | 6.5 | 15.5 | 39.9 |
Error points (original) | 1169 | 4800 | 736 |
Improved GPU time (s) | 7.2 | 16.8 | 41.9 |
Error points (improved) | 0 | 1 | 0 |