An Improved Spectroface Algorithm Based on Cubic B-Spline Wavelet Transform and Multiple Subbands Fusion

摘要

Spectroface algorithm based on wavelet transform and Fourier transform is an effective face recognition algorithm. Recognition performance must be affected because the algorithm only makes use of the lowest-frequency subband and other subbands are all discarded directly. In this paper, an improved spectroface algorithm based on cubic B-spline wavelet and multiple subbands fusion was proposed. At first, the viewpoint that detail subbands also have a lot of feature information are analyzed and verified in this paper. Then cubic B-spline wavelet is used for the improved algorithm because of its excellent capability of image edge detection and localization. The last results are given by the strategy of multiple subbands fusion. The experiment results in Yale and JAFFE face databases show that the algorithm proposed by this paper is effective, especially under the condition of illumination perturbations.