Local Multi Code Pattern Generation for Face Identification
Munikrishna Arunkumar, S. Valarmathy · Intelligent Automation & Soft Computing · 2016
AbstractThe existing LTrP method shows the relationship between the reference pixel and its neighbor’s, based on the four directional code that are computed using the first-order derivatives in both vertical and horizontal directions. It is apparent that the performance of LTrP can be improved by differentiating the edges in more than four directions. This observation has motivated us to propose a nine direction code referred to as Local Multi Code Pattern (LMCP). LMCP extracts supplementary directional information as compared with LTrP. The performance analysis of the aforementioned techniques were done with the help of ORL, Yale, Jaffee databases and also the recital analysis on the same significantly improves the recognition rate of the proposed method.