Robust multi-camera view face recognition - Computer Science > Computer Vision and Pattern RecognitionReportar como inadecuado

Robust multi-camera view face recognition - Computer Science > Computer Vision and Pattern Recognition - Descarga este documento en PDF. Documentación en PDF para descargar gratis. Disponible también para leer online.

Abstract: This paper presents multi-appearance fusion of Principal Component AnalysisPCA and generalization of Linear Discriminant Analysis LDA for multi-cameraview offline face recognition verification system. The generalization of LDAhas been extended to establish correlations between the face classes in thetransformed representation and this is called canonical covariate. The proposedsystem uses Gabor filter banks for characterization of facial features byspatial frequency, spatial locality and orientation to make compensate to thevariations of face instances occurred due to illumination, pose and facialexpression changes. Convolution of Gabor filter bank to face images producesGabor face representations with high dimensional feature vectors. PCA andcanonical covariate are then applied on the Gabor face representations toreduce the high dimensional feature spaces into low dimensional Gaboreigenfaces and Gabor canonical faces. Reduced eigenface vector and canonicalface vector are fused together using weighted mean fusion rule. Finally,support vector machines SVM have trained with augmented fused set of featuresand perform the recognition task. The system has been evaluated with UMIST facedatabase consisting of multiview faces. The experimental results demonstratethe efficiency and robustness of the proposed system for multi-view face imageswith high recognition rates. Complexity analysis of the proposed system is alsopresented at the end of the experimental results.

Autor: Dakshina Ranjan Kisku, Hunny Mehrotra, Phalguni Gupta, Jamuna Kanta Sing


Documentos relacionados