Gender Classification via Global-Local Features Fusion

摘要: Computer vision based gender classification is an interesting and challenging research topic in visual surveillance human-computer interaction systems. In this paper, on the results of psychophysics neurophysiology studies that both local global information crucial for image perception, we present effective global-local features fusion (GLFF) method classification. First, are extracted active appearance models (AAM) by LBP operator. Second, fused sequent selection Third, predicted selected via support vector machines (SVM). The experimental show proposed local-global combination scheme could significantly improve accuracy obtained either or features, leading to promising performance.