921721.fig.002a
921721.fig.002b
921721.fig.002c
Figure 2: Experiment 3: robustness to bias. In the first line from left to right are original image and three ground truth membership functions. In the second line and the third line, from left to right are original image and soft segmentations (three membership functions). The second row is the soft segmentation using Shen's model, and the third line is the soft segmentation using the proposed model.