Image Fusion Techniques for Multispectral Palm Image Enhancement

We proposed the multispectral image enhancement through image fusion by combining the data from the multiple spectrum to address the problem of accuracy and make the system robust against spoofing and to improve the accuracy of recognition, using more discriminating of palm images. Palm line features are clearer in the blue and green bands while red band can reveal some palm vein structure. The NIR band can show the palm vein structure as well as partial line information. Image fusion improves the robustness and accuracy of the palm based recognition system and attempt to combine the correlative and complementary information of multi-spectral images by pixel level fusion. Our aim is to develop a lower cost solution using fusion. We are fusing the blue band and near IR (NIR) band to enhance the palm line features and vein pattern/structure both in one image. Multispectral palm images of POLYU Database are used for the experiments. It is observed that the MIN based fusion method is preserving more edge information which is the discriminating features (palm lines) of the palm. Also visually min based method is observed to be the best out of the others.