Palm vein recognition using adaptive Gabor filter

Palm vein recognition based on multi-sampling and feature-level fusion

“…Then, the distance function is utilized to measure the similarity for authentication and recognition, but the regions to be matched must be highly aligned. As a result, the accurate palm vein ROI should be extracted based on two finger valley points first [3,4,22,23]. Because the vein texture spreads over the whole palm area, including the fingers, extracting a sub-region of the palm as the ROI will result in the loss of feature information.…”

“…Therefore, image enhancement is generally conducted before feature extraction to decrease the effect on the subsequent feature extraction and improve performance. Some related studies [3,23] adopted histogram equalization to normalize and enhance vein images; others [4,22] estimated the local mean intensity as the background illumination and then enhanced the original image by subtracting the background illumination. Considering these approaches, we presented a novel hierarchical enhancement strategy, Difference of Gaussian (DoG) and Histogram Equalization (DoG-HE), in our previous work.…”

“…Generally, vein images are collected from the finger, palm or dorsal area of the hand, with palm veins providing richer texture information, thereby improving the distinguishing ability and making them more convenient for acquisition, which is increasingly attractive to researchers. As with other single-modal biometrics, single-sample palm vein recognition has been intensively studied [2,3,4]. However, the limited biometrics information from a single biometric and single sample cannot satisfy the requirements of precision and reliability for recognition and authentication in real-world scenarios.…”

Modern Authentication Techniques in Smart Phones: Security and Usability Perspective

“…The recognition rate is very good using palm vein [45]. Experiments show that this approach is feasible and effective [46]. False acceptance rate is 0.0008% and false reject rate is 0.01% [47].…”

Palm Vein Recognition Using Scale Invariant Feature Transform with RANSAC Mismatching Removal

“…In terms of acquisition, it is possible to design contactless vein pattern acquisition systems; such systems are often more easily accepted by users as any perception of discomfort is minimized and the likelihood of cross contamination is reduced. It is also important to note that the palm is generally the most reliable hand region from which to obtain vein patterns for biometrics, as this region does not normally exhibit hair growth that may affect the quality of the images captured [3].…”