Jun-ying G et al. 2 suggested a method on fusion of FLD and 2DICA which gives best projection direction basis. The experiments are conducted on ORL, Yale and ORL damage database and results were found to be more robust. Roli et al. 3 proposed an experimental comparison between fixed and trained fusion rules for multimodal person verification. In this case, different experiments are performed and fusion was carried out for fixed and trained fusion rules. From the experimental results, fixed rules performed well under all circumstances. Many of these results fail to provide effective comparison between choices of multi-algorithmic or multimodal approaches. Mehrotra et al. 4 proposed a multi-algorithmic approach for iris, using texture and phase features. Texture features were extracted using Haar wavelets while phase features were obtained using Log-gabor wavelets. These features (texture and phase) were concatenated. M. El-Bakry et al. 5 proposed a face detection algorithm using Efficient Neural Network (ENN). To increase the performance of algorithm, they combined the two classifiers. One is based on cross correlation in frequency domain and the other is based on weights of neural net. Prabhakar et al. 6 presented a scheme for combining multiple matchers at decision level. They developed four different fingerprint verification systems, three minutiae based and one filter-based algorithm.