Investigation for face and object identification from a digital image

Abstract

The quality investigation on complex detail accuracy of an object by using the human vision in a long time is
limited on the fatigue, memory ability and skill. This research proposes the rapid simple and economical
investigation system for a complex detail object on a 2D digital image by using the wavelet transform and 2D
principal component analysis, 2DPCA. The six discriminated factors for the human face and object
identification are wavelet type, wavelet level, picture characteristic, size of the picture, number of eigenvector
and number of database. The experimental results from the suitably factor settings with the level 1 of Haarwavelet
transform, head shots, 100*100 pixel resolution size, 5 eigenvector and 8 database indicate 100 percent accuracy of the human face or object identification investigation and 98.75 percent average of every
number of eigenvectors for the human face identification.