Facial Recognition in Degraded Conditions Using Local Interest Points

摘要

In this paper we are interested in facial recognition, particularly in degraded conditions such as head pose variations, illumination, facial expressions and partial occlusions. In this context, several approaches have been used to overcome these problems and improve facial recognition. Our method consists in using a local approach based on interest points provided by the speeded-up-robust-feature descriptor for feature extraction and the k nearest neighbor combined with the k-dimensional tree for classification. The evaluation of the proposed approach on the Kinect Face DB and IST-EURECOM LFFD databases shows interesting results.