Abstract. Image Quality Assessment (IQA) is a critical part in face recognition system for helping to pick out the better quality images to assure high accuracy. In this paper, we propose a simple but efficient facial IQA algorithm based on Bayesian fusion of modified Structural Similarity (mSSIM) index and Support Vector Machine (SVM) as a reduced-reference method for facial IQA. The fusion scheme largely improves the facial IQA and consequently promotes the precision of face recognition when comparing to mSSIM or SVM alone. Experimental validation shows that the proposed algorithm works well in multiple feature spaces on many face databases.

No abstract

Image Quality Assessment (IQA) aims at automatically predicting the perceptual quality of targets with low computation complexity and high precision. However, it is usually very hard to combine all these merits into one algorithm. In this paper, we propose simple yet efficient facial image quality assessment algorithm---Reduced-Reference Automatic Ranking (RRAR) for face recognition. The RRAR contains a quality control stage and quality ranking stage based on modified structural similarity---Reduced-Reference of SSIM as the reduced reference IQA module. Experimental results show that the proposed algorithm increases the precision of face recognition with low memory consumption and computation complexity and works exceptionally well with face images captured under uncontrolled environment.

Abstract. In the field of face recognition (FR) and face image retrieval (FIR), features and metrics have received great attention in recent years due to their direct influence on the performance of a FR/FIR system. In this paper, we analyze the two factors for FIR in following steps. First, the face images are aligned to the same size, moreover, their illumination is balanced. Second, we extract classic features widely used in face recognition and retrieval, then utilize them in feature matching with different metrics. At last, face retrieval is performed based on the distances calculated with multiple metrics. We evaluate the efficiency of features and metrics by face retrieval in Face Recognition Grand Challenge (FRGC) database. Experimental results not only serves to select features and metrics for FIR, they also demonstrate that the two variables affect FR and FIR in different ways.