Fig. 4.

Visualization of the twelve features defined in Table 1. This twelve-dimensional feature space is the training set that we employed for supervised learning. The color of the points indicate the manual rating from magenta (0, poor) to green (10, good)