Extraction et fusion de points d'intérêt et textures spectraux pour l'identification, le contrôle et la sécurité

Abstract : Biometrics is an emerging technology that proposes new methods of control, identification and security. Biometric systems are often subject to risks. Face recognition is popular and several existing approaches use images in the visible spectrum. These traditional systems operating in the visible spectrum suffer from several limitations due to changes in lighting, poses and facial expressions. The methodology presented in this thesis is based on multispectral facial recognition using infrared and visible imaging, to improve the performance of facial recognition and to overcome the deficiencies of the visible spectrum. The multispectral images used in this study are obtained by fusion of visible and infrared images. The different recognition techniques are based on features extraction such as texture and points of interest by the following techniques: a hybrid feature extraction, a binary feature extraction, a similarity measure taking into account the extracted characteristics.
Complete list of metadatas

Cited literature [141 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02130804
Contributor : Abes Star <>
Submitted on : Thursday, May 16, 2019 - 10:15:09 AM
Last modification on : Monday, August 19, 2019 - 10:01:06 AM

File

85935_MAMADOU_2018_archivage.p...
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02130804, version 1

Citation

Diarra Mamadou. Extraction et fusion de points d'intérêt et textures spectraux pour l'identification, le contrôle et la sécurité. Vision par ordinateur et reconnaissance de formes [cs.CV]. Université Bourgogne Franche-Comté; Université de Cocody (Abidjan, Côte d'Ivoire), 2018. Français. ⟨NNT : 2018UBFCK031⟩. ⟨tel-02130804⟩

Share

Metrics

Record views

70

Files downloads

117