Directional Illumination Estimation Sets and Multilevel Matching Metric for Illumination-Robust Face Recognition

Cheng, Yong; Jiao, Liangbao; Tong, Ying; Li, Zuoyong; Hu, Yong; Xiao, Cao

doi:10.1109/access.2017.2766128

Cited by 18 publications

(8 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It means that only when a user's feedbacks are covered by another user, or they share a similar preference over average level, we put them into the similar user set U + , so does I + . According to this object function, we should minimize this constrained objective function with Mini-Batch Stochastic Gradient Descent (SGD), achieve the latent embedding vectors P, Q and control the learning rating using AdaGrad [30], as suggested in [31].The details are introduces in experimental section. Our training procedure is as follows:…”

Section: ) Regularization and Model Training A: Regularizationmentioning

confidence: 99%

Improving Recommendations by Embedding Multi-Entity Relationships With Latent Dual-Metric Learning

Yang

Han

et al. 2019

IEEE Access

View full text Add to dashboard Cite

Recently, latent vector embedding has become a research hotspot, with its great representative ability to measure the latent relationships among different views. However, most researches utilize the inner product of latent vectors as the representation of relationships, and they develop some embedding models based on this theory. In this paper, we take deep insight into the existing embedding models and find that utilizing the inner product may increase several problems: 1) in latent space, the inner product among three vectors may violate triangle principle; 2) the inner product cannot measure the relationships between vectors in the same category, such as user and user and item and item; and 3) the inner product cannot catch the collaborative relationships (user-user and item-item) for collaborative filtering. Along with this line, we propose a latent vector embedding model for collaborative filtering: latent dual metric embedding (LDME), which utilizes the dual-Euclidean distance in latent space, instead of the inner product, to represent different types of relationships (user-user, item-item, and user-item) with a uniform framework. Specifically, we design an embedding loss function in LDME, which can measure the close and remote relationships between entities, tackle the above problems, and achieve a more clear, well-explained embedding result. Extensive experiments are conducted on several real-world datasets (Amazon, Yelp, Taobao, and Jingdong), where the expiring results demonstrate that LDME can overperform some state-of-the-art user-item embedding models and can benefit the existing collaborative filtering models. INDEX TERMS Latent vector embedding, metric learning, collaborative filtering, recommender system.

show abstract

Section: ) Regularization and Model Training A: Regularizationmentioning

confidence: 99%

Improving Recommendations by Embedding Multi-Entity Relationships With Latent Dual-Metric Learning

Yang

Han

et al. 2019

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The primary face recognition discoveries under changing light variations were based on grayscale face images. Several previous studies presented methods to compensate effects of diverse illumination by utilizing grayscale images either directly or converting color images in grayscale form [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34]. However, the potency of color information for digital images with diverse environmental circumstances is necessary to gain selective discriminative facial features for efficient color face recognition.…”

Section: Introductionmentioning

confidence: 99%

Low Frequency DD-DWT based illumination compensated and color intensified features for Face Recognition

Chaudhary

Yadav

2023

Preprint

View full text Add to dashboard Cite

The efficacy of face recognition systems is significantly affected by uneven light deviations, incident over images from different directions. This work presents a light invariable color face recognition method that efficiently normalizes illumination variances along with substantial improvement in color intensities. The proposed method includes following steps: Initially, segregation of facial images into primary spectral color components is performed to effectively balance contrast and evenly amplify color intensity levels. The varying illumination effects are due to low frequency component of image; therefore, Difference of Gaussian (DoG) high-pass filter based homomorphic filtering is further utilized in HSI (Hue, Saturation and Intensity) space. This conversion into HSI color space is mainly performed to ward off any colors distortion that may prevail in images due to direct utilization of further improvements on each color channel. Thus, this model aids in easily separating the chromaticity component from intensity part of an image. The normalized intensity component is concatenated with unaltered chromaticity components to effectively possess color facial information. Subsequently, extraction of double density discrete wavelet transform (DD-DWT) based coefficients of images is achieved for selection of substantial discriminative frequency components of face images. The large facial feature vector space is hereafter projected over eigen subspace to reduce dimensions effectively. In this work, k-nearest neighbor linear classifier has been utilized to classify selective features. The effectiveness of investigated method has been assessed on AR and CMU-PIE color face databases and attained results are thus contrasted with prior techniques which benefit in establishing its superiority.

show abstract

“…Cheng et al introduced a multi-scale principal contour direction method (MPCD) [25] through studying multi-scale description of real-world images, contour features of face images and feature fusion techniques. A directional illumination estimation set algorithm [26] was proposed to improve the MPCD method, which greatly alleviates the false contours of shadow edges and preserves the discriminative characteristics of the real-world face images. A feature extraction method called R1-norm2-D principal component analysis (R1-2-DPCA) was presented in [27].…”

Section: Introductionmentioning

confidence: 99%

Robust face recognition for occluded real‐world images using constrained probabilistic sparse network

et al. 2022

IET Image Processing

View full text Add to dashboard Cite

Aiming at the occluded real-world face images across illumination, pose, expression, and resolution variations, a robust face recognition for occluded real-world images using constrained probabilistic sparse network is presented. A constrained probabilistic sparse representation network is constructed to obtain the features of all the training images from a global perspective, and the new network nodes are generated through the random combination of the training images. In the probabilistic sparse representation network, the probabilities of each class of the sparse subspace that the occluded test images individually belong to are defined and calculated. The final classifications of the test images are determined by the joint maximum probability of the network nodes. Meanwhile the second-order gradient constraint is the first introduced in the probabilistic sparse representation network. It is found that the constraint uses the adjacent pixels of the face images to obtain the local texture similarity, and further use the local texture similarity to distinguish the occlusion and non-occlusion parts. Thus the constraint can reduce the influence of the occlusion part on face recognition. Extensive experiments with the 12 existing methods on the five face databases demonstrate that the recognition rate of the proposed method is the best than the non-deep learning methods compared, and the proposed method can obtain nearly the same recognition rate with an advantage of a very less time consumption compared to the state-of-the-art deep learning methods.

show abstract

Directional Illumination Estimation Sets and Multilevel Matching Metric for Illumination-Robust Face Recognition

Cited by 18 publications

References 39 publications

Improving Recommendations by Embedding Multi-Entity Relationships With Latent Dual-Metric Learning

Improving Recommendations by Embedding Multi-Entity Relationships With Latent Dual-Metric Learning

Low Frequency DD-DWT based illumination compensated and color intensified features for Face Recognition

Robust face recognition for occluded real‐world images using constrained probabilistic sparse network

Contact Info

Product

Resources

About