Impact of Face Registration Errors on Recognition

Rentzeperis, Ilias; Stergiou, Andreas; Pnevmatikakis, Aristodemos; Polymenakos, Lazaros

doi:10.1007/0-387-34224-9_22

Cited by 31 publications

(34 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The second difference is, the manual eye center labels, which have been used for alignment is more precise in the artificially occluded face images, whereas due to sunglasses, manual eye center labels of the face images from the AR face database is not reliable. Recent studies on the impact of registration errors on face recognition performance [16,17] lead us to hypothesize that the second difference has more dominant effect. By utilizing an alignment approach [22], which is insensitive to imprecise facial feature localization, we run the experiments again and obtained very high increase in the performance indicating that the main reason causing the performance drop is not the missing eye region information, but the misalignment due to erroneously localized facial feature points.…”

Section: Discussionmentioning

confidence: 99%

“…This implies that, there is another underlying problem, in addition to missing eye region information. Recently, some studies have been conducted on the robustness of face recognition algorithms against registration errors [16,17]. These studies have shown that face recognition algorithms' performance rely on face alignment accuracy.…”

Section: Ar1scarfmentioning

confidence: 99%

“…It is found that, the eigenfaces approach can tolerate variations only upto 5% of the face image size. Similarly in [17], in addition to the eigenfaces approach, Fisherfaces [10], elastic bunch graph matching [18], Laplacianfaces [19], minimum average correlation energy (MACE) [20], and pseudo two-dimensional hidden Markov models [21] have been analyzed against registration errors. All these algorithms are observed to be sensitive to misalignment.…”

Section: Ar1scarfmentioning

confidence: 99%

See 2 more Smart Citations

Why Is Facial Occlusion a Challenging Problem?

Ekenel¹,

Stiefelhagen²

2009

Advances in Biometrics

View full text Add to dashboard Cite

Abstract. This paper investigates the main reason for the obtained low performance when the face recognition algorithms are tested on partially occluded face images. It has been observed that in the case of upper face occlusion, missing discriminative information due to occlusion only accounts for a very small part of the performance drop. The main factor is found to be the registration errors due to erroneous facial feature localization. It has been shown that by solving the misalignment problem, very high correct recognition rates can be achieved with a generic local appearance-based face recognition algorithm. In the case of a lower face occlusion, only a slight decrease in the performance is observed, when a local appearance-based face representation approach is used. This indicates the importance of local processing when dealing with partial face occlusion. Moreover, improved alignment increases the correct recognition rate also in the experiments against the lower face occlusion, which shows that face registration plays a key role on face recognition performance.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Ar1scarfmentioning

confidence: 99%

Section: Ar1scarfmentioning

confidence: 99%

See 1 more Smart Citation

Why Is Facial Occlusion a Challenging Problem?

Ekenel¹,

Stiefelhagen²

2009

Advances in Biometrics

View full text Add to dashboard Cite

show abstract

“…As most generic face recognition approaches do not explicitly model local deformations, face registration errors can have a large impact on the system performance [18]. In contrast to the manually aligned database setup, a second database setup has been generated where the faces have been automatically detected using the OpenCV implementa- Modular PCA [22] 14.14 Adaptively weighted Sub-Pattern PCA [22] 6.43 DCT [4] 4.70 tion of the Viola & Jones [23] face detector.…”

Section: Conditions and Resultsmentioning

confidence: 99%

SURF-Face: Face Recognition Under Viewpoint Consistency Constraints

Dreuw

Steingrube

Hanselmann

et al. 2009

Procedings of the British Machine Vision Conference 2009

View full text Add to dashboard Cite

We analyze the usage of Speeded Up Robust Features (SURF) as local descriptors for face recognition. The effect of different feature extraction and viewpoint consistency constrained matching approaches are analyzed. Furthermore, a RANSAC based outlier removal for system combination is proposed. The proposed approach allows to match faces under partial occlusions, and even if they are not perfectly aligned or illuminated.Current approaches are sensitive to registration errors and usually rely on a very good initial alignment and illumination of the faces to be recognized.A grid-based and dense extraction of local features in combination with a block-based matching accounting for different viewpoint constraints is proposed, as interest-point based feature extraction approaches for face recognition often fail.The proposed SURF descriptors are compared to SIFT descriptors. Experimental results on the AR-Face and CMU-PIE database using manually aligned faces, unaligned faces, and partially occluded faces show that the proposed approach is robust and can outperform current generic approaches.

show abstract

“…And the man-made variations are mainly due to the imperfections of the capture devices and image processing technologies, e.g. the noises from the cameras and the face registration error resulting from imperfect face detections [1]. The performances of many recognition algorithms degrade significantly in these cases.…”

Section: Introductionmentioning

confidence: 99%

Feature Correlation Filter for Face Recognition

Zhu

Liao

Lei

et al.

Advances in Biometrics

View full text Add to dashboard Cite

Abstract. The correlation filters for pattern recognition, have been extensively studied in the areas of automatic target recognition(ATR) and biometrics. Whereas the conventional correlation filters perform directly on image pixels, in this paper, we propose a novel method, called "feature correlation filter (FCF)", by extending the concept of correlation filter to feature spaces. The FCF preserves the benefits of conventional correlation filters, i.e., shift-invariant, occlusion-insensitive, and closed-form solution, and also inherits virtues of the feature representations. Moreover, since the size of feature is often much smaller than the size of image, the FCF method can significantly reduce the storage requirement in recognition system. The comparative results on CMU-PIE and the FRGC2.0 database show that the proposed FCFs can achieve noteworthy performance improvement compared with their conventional counterpart.

show abstract

Impact of Face Registration Errors on Recognition

Cited by 31 publications

References 13 publications

Why Is Facial Occlusion a Challenging Problem?

Why Is Facial Occlusion a Challenging Problem?

SURF-Face: Face Recognition Under Viewpoint Consistency Constraints

Feature Correlation Filter for Face Recognition

Contact Info

Product

Resources

About