Multi-modal palm-print and hand-vein biometric recognition at sensor level fusion

Abstract

When it is important to authenticate a person based on his or her biometric qualities, most systems use a single modality (e.g. fingerprint or palm print) for further analysis at higher levels. Rather than using higher levels, this research recommends using two biometric features at the sensor level. The Log-Gabor filter is used to extract features and, as a result, recognize the pattern, because the data acquired from images is sampled at various spacing. Using the two fused modalities, the suggested system attained greater accuracy. Principal component analysis was performed to reduce the dimensionality of the data. To get the optimum performance between the two classifiers, fusion was performed at the sensor level utilizing different classifiers, including k-nearest neighbors and support vector machines. The technology collects palm prints and veins from sensors and combines them into consolidated images that take up less disk space. The amount of memory needed to store such photos has been lowered. The amount of memory is determined by the number of modalities fused.