Manifold based analysis of facial expression

Chang, Yu-Hao; Hu, Chaofang; Feris, Rogerio; Turk, Matthew

doi:10.1016/j.imavis.2005.08.006

Cited by 167 publications

(74 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…or appearance features representing the texture of the facial skin including wrinkles, bulges, and furrows. Typical examples of geometric-feature-based methods are those of Gokturk et al (2002), who used 19 point face mesh, of Chang et al (2006), who used a shape model defined by 58 facial landmarks, and of Pantic and her collegues (Pantic & Rothkrantz, 2004;Pantic & Patras, 2006;Valstar & Pantic, 2006a), who used a set of facial characteristic points like the ones illustrated in Figure 3. Typical examples of hybrid, geometric-and appearance-feature-based methods are those of Tian et al (2001), who used shape-based models of eyes, eyebrows and mouth and transient features like crows-feet wrinkles and nasolabial furrow, and of Zhang and Ji (2005), who used 26 facial points around the eyes, eyebrows, and mouth and the same transient features as Tian et al (2001).…”

Section: Facial Feature Extractionmentioning

confidence: 99%

Machine Analysis of Facial Expressions

Pantić¹,

Bartlett²

2007

Face Recognition

176

147

View full text Add to dashboard Cite

Section: Facial Feature Extractionmentioning

confidence: 99%

Machine Analysis of Facial Expressions

Pantić¹,

Bartlett²

2007

Face Recognition

176

147

View full text Add to dashboard Cite

“…The recognition rate is depicted in a graph (Figure 9(b)) according to the expressions. The proposed work produced the highest recognition rate when compared to existing methods like Active Appearance Model [31], Scaled Gaussian Process Regression (SGPR) [8], Coupled SGPR (CSGPR) [4] and the combination of SVM [32] and HM [16].…”

Section: Performance Testingmentioning

confidence: 99%

Recognizing Expression Variant and Occluded Face Images Based on Nested HMM and Fuzzy Rule Based Approach

Ramalingam¹,

Dhanushkodi²

2016

View full text Add to dashboard Cite

The face recognition with expression and occlusion variation becomes the greatest challenge in biometric applications to recognize people. The proposed work concentrates on recognizing occlusion and seven kinds of expression variations such as neutral, surprise, happy, sad, fear, disgust and angry. During enrollment process, principle component analysis (PCA) detects facial regions on the input image. The detected facial region is converted into fuzzy domain data to make decision during recognition process. The Haar wavelet transform extracts features from the detected facial regions. The Nested Hidden markov model is employed to train these features and each feature of face image is considered as states in a Markov chain to perform learning among the features. The maximum likelihood for the input image was estimated by using Baum Welch algorithm and these features were kept on database. During recognition process, the expression and occlusion varied face image is taken as the test image and maximum likelihood for test image is found by following same procedure done in enrollment process. The matching score between maximum likelihood of input image and test image is computed and it is utilized by fuzzy rule based method to decide whether the test image belongs to authorized or unauthorized. The proposed work was tested among several expression varied and occluded face images of JAFFE and AR datasets respectively.

show abstract

“…Semi-supervised and transfer learning: Manifold learning has been widely explored in computer vision, such as face expression [8] and age estimation [13]. The intuition of incorporating manifold regularisation in semi-supervised learning has also been studied [1,4,38,18], whilst manifold-based transfer learning has been proposed in [34] to transfer knowledge across domains via an aligned manifold.…”

Section: Related Workmentioning

confidence: 99%