Developing Image Processing Meta-Algorithms with Data Mining of Multiple Metrics

Figure 3

The table of metric values shown in Figure 1, after replacing metric values by their rankings, can be analyzed with principal component analysis (PCA). Replacement by rankings yields what is known as robust PCA, a nonparametric approach to dimensionality reduction with reduced sensitivity to outliers. The 11-dimensional metric value dataset is reduced here to a 2-dimensional plot along the first two principal components, showing that the FLIRT results (red points) generally dominate the others along the first principal component (𝑥-axis). The AIR Warp results (green points) can dominate if we change the metric emphasis to the second principal component (𝑦-axis).