Design of contactless hand biometric system with relative geometric parameters

The main concern in contactless biometric system is the position of the hand that may vary relatively to camera. This variation of position result in different parameters of hand geometry captured from the same person, in different time, which is a serious problem in identification process. This paper presents a novel contactless biometric verification system based on relative geometric parameters of the hand, as the biometric feature. A webcam captured color image of the hand, which will be transformed into binary image for segmentation, based on thresholding technique. Binary image was extracted to get nine absolute geometrical sizes. The relative geometric parameters derived from ratios between those geometrical sizes were calculated. Feature extraction was processed by scanning technique. Matching was conducted based on match score, which is the output result by feeding relative geometric parameters to backpropagation-trained artificial neural network. Our design provided accuracy of 87.237%, precision of 85.798%, False Match Rate (FMR) of 14.780%, and False Non Match Rate (FNMR) of 10.747%.