Skip to main content
. 2018 Jul 6;9(8):3590–3606. doi: 10.1364/BOE.9.003590

Fig. 11.

Fig. 11

(Best viewed in color) Results for skin surface segmentation for the proposed approach as compared with related works. From left to right are ground truth and the results of A. Li et al. [21], K. Li et al. [12] and the proposed method, respectively. The unsigned mean vertical errors for each sample (from left to right respectively) are (i) 3.24, 3.46 and 2.38 pixels (ii) 11.45, 4.61 and 3.67 pixels.