Selecting best spectrum using multispectral palm texture

Abstract

Multispectral palm print is one of the most reliable and unique Biometric. MSI have faster acquisition time and better quality images than normal images. The advantages of the proposed method include better hygiene and higher verification performance. In this we proposed Local binary Pattern (LBP) based histogram for multispectral palm print representation and to choose the best spectrum for authentication. Here the central part of the palm print image is resized to the size of 180 × 180 and divided into non overlapping sub-images. The size of the sub-image various from 2×2 pixels to 90×90 pixels. The histogram is obtained for each block and the values are used for comparison. Totally 36 images per person are taken from standard database available. Training set is prepared with the help of 2 images from each spectrum. Results are checked against remaining images in authentication mode. Results are represented in terms of Genuine acceptance rate(%). Most of the palm print recognition systems use white light to acquire Images. This study analyzes the palm print recognition performance under six different illuminations, including the white light. The experimental results with a large database show that white light is not the optimal illumination, while 700nm light could achieve higher palm print recognition accuracy than the white light. In authentication mode 98% recognition rate is obtained for the spectrum 700nm. The experiment was conducted for six spectrums like 460,630,700,850,940nm, White Light. We use the CASIA-MS-Palmprint V1 database of size 7200 images collected by the Chinese Academy of Sciences' Institute of Automation (CASIA).