This method [1] encode the grayscale image by comparing selected pixel value to its neighbour. The comparing result will be encoded into binary  value as follows:

At the end, the encoded binary value will be converted back to decimal value. Those decimal value will be used as feature vectors.

Figure 1: LBP process

Bibliography

[1]  T. Ojala, M. Pietikainen, and T. Maenpaa.  Multiresolution gray-scale    and rotation invariant texture classification with local binary patterns. IEEE Transactions on Pattern Analysis and Machine Intelligence, 24(7): 971–987, Jul 2002. ISSN 0162-8828. doi: 10.1109/TPAMI.2002.1017623.