Real Time Face Recognition Using Polynomial Regression and Sub-region Color Component Distribution

An efficient architecture for real time face recognition is presented here using Polynomial Regression for feature edge detection and by determining color-component distribution for feature-regions. Here we determine second order polynomial equation by polynomial regression for the edges of eyelid and chin. Chin does not change in different expressions, the change of eyelid is also rare and these show clear edges in the picture. For determining polynomial equations the coordinate system is very important here. The same curve may have different equations depending on the position of the curve on the graph. We have eliminated this problem. As we derive second-order polynomial equations we have three constants for each curve which we can call A, B and C. We also determine red, green and blue color-distribution values for three regions eye-region, lipregion and nose-region. For the training values related to the same person these values are averaged. Finally the recognition is performed based on the weighted sum of errors obtained from A, B, C values of the edges and color distribution values of the regions. This method is too much faster and its recognition efficiency is high