Research Article

An Efficient and Robust Iris Segmentation Algorithm Using Deep Learning

Figure 8

Illustration of boundary point selection algorithm. (a) 11 emitting paths are drawn with the parameter (for the other side ), and the points corresponding to the maximal gradient are recorded. The reference value is determined by taking the median of the distances from these points to the pupillary center. (b–e) Repeatedly drawing new emitting paths and applying the boundary point selection algorithm result in the correct location of many boundary points. (f) Final limbus boundary is recovered by fitting a circle on all candidate points based on Equation (4).
(a)
(b)
(c)
(d)
(e)
(f)