Likelihood ratio based mixed resolution facial comparison

Peng, Yuhao; Spreeuwers, Luuk; Veldhuis, Raymond

doi:10.1109/iwbf.2015.7110231

Cited by 10 publications

(29 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…After training with high-resolution and low-resolution image pairs, the MixRes classifier can directly compare lowresolution probe to high-resolution gallery. It is shown in [5] that this method has promising performance on very lowresolution probes.…”

Section: Scenario and Hypothesismentioning

confidence: 97%

“…To solve the first problem, we use the MixRes classifier [5]. After training with high-resolution and low-resolution image pairs, the MixRes classifier can directly compare lowresolution probe to high-resolution gallery.…”

Section: Scenario and Hypothesismentioning

confidence: 99%

“…Here we briefly describe the MixRes classifier, which is originally named Mixed-Resolution Biometric Comparison [5]. The MixRes classifier is especially designed for comparing images captured at different distances.…”

Section: Mixres Classifiermentioning

confidence: 99%

“…They are built up of a PCA step, reducing the dimensionality of probe and reference to manageable proportions as well as whitening them, followed by an LDA step that aims for optimal discrimination after transformation to a common lower dimensional subspace. This method and the training procedure are described in [5]. The log-likelihood ratio is then computed as…”

Section: Mixres Classifiermentioning

confidence: 99%

“…The mappings aim at increasing class-separation for highresolution images and mapping low-resolution images to their corresponding class-separated high-resolution data. Peng et al [5] proposed a likelihood ratio based method for direct comparison between images of different resolutions. All these works consider a specific resolution.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Designing a Low-Resolution Face Recognition System for Long-Range Surveillance

Peng

Spreeuwers

Veldhuis

2016

2016 International Conference of the Biometrics Special Interest Group (BIOSIG)

View full text Add to dashboard Cite

Most face recognition systems deal well with highresolution facial images, but perform much worse on lowresolution facial images. In low-resolution face recognition, there is a specific but realistic surveillance scenario: a surveillance camera monitoring a large area. In this scenario, usually the gallery images are of high-resolution and the probe images are of various low-resolutions depending on the distances between the subject and the camera. In this paper, we design a low-resolution face recognition system for this scenario. We use a state-ofthe-art mixed-resolution classifier to deal with the resolution mismatch between the gallery and probe images. We also set up experiments to explore the best training configuration for probe images of various resolutions. Our experimental results show that one classifier which is trained on images of various resolutions covering the whole range has promising results in the long-range surveillance scenario. This system has at least as good performance as combining multiple face recognition systems that are optimised for different resolutions.

show abstract

Section: Scenario and Hypothesismentioning

confidence: 97%

Section: Scenario and Hypothesismentioning

confidence: 99%

Section: Mixres Classifiermentioning

confidence: 99%

Section: Mixres Classifiermentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Designing a Low-Resolution Face Recognition System for Long-Range Surveillance

Peng

Spreeuwers

Veldhuis

2016

2016 International Conference of the Biometrics Special Interest Group (BIOSIG)

View full text Add to dashboard Cite

show abstract

Low‐resolution face alignment and recognition using mixed‐resolution classifiers

2017

View full text Add to dashboard Cite

A very common case for law enforcement is recognition of suspects from a long distance or in a crowd. This is an important application for low-resolution face recognition (in the authors' case, face region below 40 × 40 pixels in size). Normally, high-resolution images of the suspects are used as references, which will lead to a resolution mismatch of the target and reference images since the target images are usually taken at a long distance and are of low resolution. Most existing methods that are designed to match high-resolution images cannot handle low-resolution probes well. In this study, they propose a novel method especially designed to compare low-resolution images with high-resolution ones, which is based on the log-likelihood ratio (LLR). In addition, they demonstrate the difference in recognition performance between real low-resolution images and images down-sampled from high-resolution ones. Misalignment is one of the most important issues in low-resolution face recognition. Two approaches-matching-score-based registration and extended training of images with various alignmentsare introduced to handle the alignment problem. Their experiments on real low-resolution face databases show that their methods outperform the state-of-the-art.

show abstract

Low‐resolution face recognition and the importance of proper alignment

2019

View full text Add to dashboard Cite

Face recognition methods for low resolution are often developed and tested on down-sampled images instead of on real low-resolution images. Although there is a growing awareness that down-sampled and real low-resolution images are different, few efforts have been made to analyse the differences in recognition performance. Here, the authors explore the differences and demonstrate that alignment is a major cause, especially in the absence of pose and illumination variations. The authors found that the recognition performances on down-sampled images are flattered mostly due to the fact that the images are perfectly aligned before down-sampling using high-resolution landmarks, while the real low-resolution images have much poorer alignment. To obtain better alignment for real low-resolution images, the authors apply matching score-based registration which does not rely on accurate landmarks. The authors propose to divide low resolution into three ranges to harmonise the terminology: upper low resolution (ULR), moderately low resolution (MLR), and very low resolution (VLR). Most face recognition methods perform well on ULR. MLR is a challenge for commercial systems, but a low-resolution deep-learning method can handle it very well. The performance of most methods degrades significantly for VLR, except for simple holistic methods which perform the best.

show abstract

Likelihood ratio based mixed resolution facial comparison

Cited by 10 publications

References 14 publications

Designing a Low-Resolution Face Recognition System for Long-Range Surveillance

Designing a Low-Resolution Face Recognition System for Long-Range Surveillance

Low‐resolution face alignment and recognition using mixed‐resolution classifiers

Low‐resolution face recognition and the importance of proper alignment

Contact Info

Product

Resources

About