Table 3.
List of the 20 most robust features over the two vendors examined in this study. The composite indicator is a measure of robustness, where larger values indicate a more robust feature relative to the others examined in this study. The composite indicator is computed according to Eq. (1)
| Feature name | Feature family | Composite indicator (CI) |
|---|---|---|
| Sum entropy | GLCM | 5.81 |
| Percentage density | Density | 5.52 |
| Dim 5 | Box‐counting fractal dimension | 5.33 |
| Sum variance | GLCM | 5.26 |
| Beta 3 | Power law | 5.23 |
| safmp | Fourier features | 5.18 |
| Beta 1 | Power law | 5.18 |
| Variance | GLCM | 5.14 |
| Dim 4 | Box‐counting fractal dimension | 5.11 |
| Beta 7 | Power law | 5.09 |
| IMC 2 | GLCM | 5.06 |
| Maximum correlation coefficient | GLCM | 5.01 |
| Dim 1 | Box‐counting fractal dimension | 4.99 |
| Correlation | GLCM | 4.96 |
| Sarms | Fourier features | 4.91 |
| Beta 5 | Power law | 4.81 |
| rrms | Fourier features | 4.80 |
| rfmp | Fourier features | 4.72 |
| Global Minkowski dimension | Minkowski fractal dimension | 4.71 |
| Dim | Box‐counting fractal dimension | 4.66 |