Optimized Periocular Template Selection for Human Recognition

Algorithm 2

Open_Eye_Detection.

Require: 𝐼: RGB face image of size 𝑚×𝑛, 𝑆: Binary face image indicating skin-map

Ensure: EM: Binary face image indicating open eye regions

 (1) Convert RGB image 𝐼 to 𝑌𝐶_𝑏𝐶_𝑟 color space

 (2) Normalize _𝐼𝑌_𝑖,𝑗 to [0,255] where _𝐼𝑌_𝑖,𝑗 denotes 𝑌 value for the pixel (𝑖,𝑗)

 (3) FC = Set of connected components in 𝑆

 (4) EM=FC

 (5) For each connected component EM_𝑝 in 𝑆, repeat Step 5 to 8

 (6) EB_𝑝=0

 (7) For each pixel (𝑖,𝑗) in EM_𝑝, the value of EB_𝑝 is updated as,

  EB_𝑝=EB_𝑝+0,if65<_𝐼𝑌_𝑖,𝑗<80EB_𝑝+1,otherwise

 (8) if EB_𝑝 = Number of pixels in EM_𝑝, then do EM=EM−EM_𝑝

  (Removal of the 𝑝th connected component)

 (9) return EM