Component-Based Recognition of Facesand Facial Expressions

Abstract

Abstract-Most of the existing methods for the recognition of faces and expressions consider either the expression-invariant face recognition problem or the identity-independent facial expression recognition problem. In this paper, we propose joint face and facial expression recognition using a dictionary-based component separation algorithm (DCS). In this approach, the given expressive face is viewed as a superposition of a neutral face component with a facial expression component which is sparse with respect to the whole image. This assumption leads to a dictionary-based component separation algorithm which benefits from the idea of sparsity and morphological diversity. This entails building data-driven dictionaries for neutral and expressive components. The DCS algorithm then uses these dictionaries to decompose an expressive test face into its constituent components. The sparse codes we obtain as a result of this decomposition are then used for joint face and expression recognition. Experiments on publicly available expression and face datasets show the effectiveness of our method.

Discriminant Incoherent Component Analysis

“…For RPCA+SRC, RPCA [9] is applied for each subject and the resulting low-rank (sparse) matrices are used for SRC-based face (expression) recognition similarly to [11]. For LRSI, the algorithm in [12] is applied subjectwise for face recognition and expression-wise for expression recognition; the nuclear norm is used for all components.…”

“…Choosing an initial estimate that is close to the optimum sought can markedly speed up the convergence of a non-convex optimization problem like the DICA [47]. RPCA has been proved efficient in recovering low-complexity facial components, while also being robust to gross errors in the data [11]. This motivates its choice for the initialization step, while its positive impact on the convergence speed was corroborated by preliminary experiments.…”

“…Finally, the component separation approach of [11] is followed, where the reconstruction imageŷ I = D IαI based on the identity dictionary D I is utilized for face recognition, and, similarly, the reconstruction imageŷ E = D EαE based on the expression dictionary D E is utilized for expression recognition, according to the following minimum-residual rules:…”

“…Another SVD-based work is [37], where the proposed HigherOrder SVD is used to learn the mapping between persons and expressions, which is subsequently utilized to perform facial expression decomposition. Recently, Taheri et al [11] combine RPCA [9] and K-SVD [38] to construct one identity and one expression dictionary, which are in turn fed into a SRC-like framework for joint face and expression recognition.…”

Scrambling Face Images for Privacy Protection Using Arnold Transform.

“…Taheri et al [14]. proposes Component-based recognition of faces and facial expressions using a dictionary-based component separation algorithm.…”