Layered Local Binary Pattern for Face Detection

Interaction between man and machine can be obtained by detecting and classifying faces. The challenges for practical face detection and categorization are fewer lighting conditions and noise. LBP is the best technique for face classification and detection because LBP is a non-parametric descriptor that capably summarizes the local structure of images. Due to the dependability of LBP, various LBP techniques are established as well as demonstrated for the improvement of face recognition performance. This technique is insensitive to noise and low contrasted faces. However, this technique is failed to retain small important features like stickers on the forehead and moles on the face which carry useful information. To overcome these drawbacks, the layered local binary pattern is proposed in this project. For computational convenience, the eight binary codes are transformed at each layer into a decimal number.