| Summary: | 碩士 === 國立成功大學 === 電機工程學系碩博士班 === 100 === Face recognition has been an active research area due to its wide range of application in information security, law enforcement, human-computer interaction, airport security, and video surveillance systems. Face recognition commits errors such as illumination conditions, pose, facial expression, aging, partial occlusions.
In the thesis, we propose global and local discriminative features for face recognition under various facial conditions. The global and local discriminative features consists of two mainly discriminant information for face recognition. The global feature extracted from the whole face image using local linear discrimination analysis. The local feature selects four facial features using local binary patterns and local linear discrimination analysis. Local linear discriminant analysis can preserve the information of the region structure. The local classifier combines classifier of four local regions. The combination classifier combines the global and local classifiers.
As shown in experimental result, the global and local discriminative features can have more effective discriminative power than traditional face recognition methods.
|
|---|
