Upgrading the precision of face recognition using the gradient of a facet function
Hee Sung Kim, Jong-Ho Kim · 2006
One of the most difficult limitations in face recognition by computer vision process is to treat the shadows on human face which vary according to the direction of the light. To enhance the rate of correct recognition overcoming this problem, we apply a facet function on each local surface of the face image. If the image is appropriately divided into the same size of the local surfaces, the gradient of the facet function representing each local surface can be calculated. The set of magnitude and direction of the gradient reflects the inherent shapes of a face image. The size of the local surface that we used for the computation of gradient is 5 /spl times/ 5. The recognition rate of face using this method for our sample images is computed to be 96.5 and it is shown that this rate is the best comparing to the ones produced by the existent methods.