Fusion of Image Feature Descriptors for Person Re-identification

Abstract

Person re-identification has gained a lot of research interest in recent years. Extracting and matching features play an important role in this scenario. Past studies of image feature detectors and descriptors are more generic in nature. Different types of detectors and descriptors are used for person re-identification over the last few years. Most of these descriptors are a combination of two or more variants of descriptors. This research paper will focus on the comparative analysis and evaluation of various features detectors and descriptors used for image matching with relevance to person re-identification. We also explore how the combination of local and global descriptors can improve the re-identification rate. VIPeR dataset is used for the evaluation of descriptors.