External Feature Extraction of Eyebrows

To study the external feature extraction method of eyebrows, the pseudo-sphere-based edge detector and Li's model are integrated, the eyebrow contour of the pure eyebrow image is obtained by level sets evolution, features of shape, direction and texture of the eyebrow are calculated, and the eyebrow external features model is built. Experimental results show that the eyebrow contour from the proposed method is more accurate than that from Li's model after the same iterations. To a self-built nature eyebrow images library (100 samples) and with external features model, the recognition rate of the single eyebrow can reach 86.1%, similar to those from HMM and 2DPCA, and 90.2% of the double eyebrows. Even to the eyebrows library, in which eyebrows have no difference in heavy-light, the model with only shape and direction features is also valid, and the recognition rate can reach 88.1%.