Deep Representation for Finger-Vein Image-Quality Assessment

A Systematic Review of Finger Vein Recognition Techniques

“…Experimental results of Reference [121] outperform some traditional conventional finger vein recognition approaches. Moreover, Qin et al [122] have successfully applied a deep learning algorithm to assess the quality of finger vein image and have achieved higher identification accuracy with respect to current traditional state-of-the-art image quality assessment. In Reference [123], a deep convolutional neural network (DCNN) with hard mining finger verification method was proposed which achieved better performance than commercial finger vein verification methods.…”

“…Feature extraction is one of the main steps in FVR. Deep learning approaches are robust to learn features directly from raw pixels, without the need for handcrafted descriptors, which greatly improves matching performance [122]. However, in conventional approaches handcrafted descriptors (Curvature, Gabor filter, Radon transform, Information capacity, etc.)…”

Finger-Vein Quality Assessment Based on Deep Features From Grayscale and Binary Images

“…We have set, in this regard, a new DNN framework that does not require manual labeling of high and low quality images as it does by state of the art methods, but infers such annotations automatically based on an objective indicator, the veri¯cation decision output. This framework has signi¯cantly outperformed the existing methods, whether the DNN input image is in grayscale 16 or is binary. 18 Motivated by these performance levels, we propose, in this work, a representation learning of¯nger vein image quality, where a DNN takes as input conjointly the grayscale and binary versions of the input image to predict its quality.…”

“…The quality vein attributes are therein determined by human intuition and a priori knowledge as they are detected by hand-crafted descriptors. As a result, they su®er from the following drawbacks 16 : First, it is di±cult to demonstrate that the manually selected attributes are related to image quality. Second, it is impossible to investigate all the attributes a®ecting image quality.…”

“…Recently, we have proposed the¯rst Deep Neural Network (DNN) for assessinḡ nger-vein quality, 16 and actually for any biometrics, based on very limited knowledge. The main issue for applying Deep Learning, in the image quality estimation context, was the inavailability of annotations (labels) on the input quality.…”