Personal identification in contact free environment by multispectral palm image fusion

In this paper, we propose a contact free system for palm image recognition. Palm images in Red, Green and Blue spectra are used. The method involves multispectral image acquisition using a dedicated setup, pre processing to dynamically locate the region of interest and image fusion using Fourier and wavelet transform based techniques. In pre processing movement of palm in one plane is taken care of by dynamically locating the region of interest. We used modified fusion criterion in Fourier domain to give more emphasis on Blue images. The tests were carried out on 450 multispectral images of 50 people. Wavelet based technique is found to be superior because it shows more difference in matching parameter between matched and unmatched image pairs.