Face recognition using discrete haar wavelet transforms

Abstract

Wavelet transform is being applied as an important analyzing tool in signal and image processing because it provides better signal representations with less computation time because of property of multiresolution analysis. An analogue image is transformed into a discrete image in a space through a sampling process. The wavelet transforms and analysis of discrete image are performed using Haar wavelet, level-2. Through analyzing and comparing different patterns, one or more persons are identified in the face recognition process. It is accomplished by using correlation coefficient and Euclidean distance of scaling coefficients of wavelet transformed images. In the face recognition the typically facial features are extracted and compared to a database for finding the best match. Matching of features is performed between test image and stored database images.