Person re-identification via rich color-gradient feature | IEEE Conference Publication | IEEE Xplore

Person re-identification via rich color-gradient feature


Abstract:

Person re-identification refers to match the same pedestrian across disjoint views in non-overlapping camera networks. Lots of local and global features in the literature...Show More

Abstract:

Person re-identification refers to match the same pedestrian across disjoint views in non-overlapping camera networks. Lots of local and global features in the literature are put forward to solve the matching problem, where color feature is robust to viewpoint variance and gradient feature provides a rich representation robust to illumination change. However, how to effectively combine the color and gradient features is an open problem. In this paper, to effectively leverage the color-gradient property in multiple color spaces, we propose a novel Second Order Histogram feature (SOH) for person reidentification in large surveillance dataset. Firstly, we utilize discrete encoding to transform commonly used color space into Encoding Color Space (ECS), and calculate the statistical gradient features on each color channel. Then, a second order statistical distribution is calculated on each cell map with a spatial partition. In this way, the proposed SOH feature effectively leverages the statistical property of gradient and color as well as reduces the redundant information. Finally, a metric learned by KISSME [1] with Mahalanobis distance is used for person matching. Experimental results on three public datasets, VIPeR, CAVIAR and CUHK01, show the promise of the proposed approach.
Date of Conference: 11-15 July 2016
Date Added to IEEE Xplore: 29 August 2016
ISBN Information:
Electronic ISSN: 1945-788X
Conference Location: Seattle, WA, USA

1. Introduction

In video surveillance, it is desirable to judge whether or not a pedestrian has appeared across disjoint camera views. That is person re-identification problem. It has received increasing attention recently as it could greatly save human effort on manually browsing and searching persons in a large scale dataset. However, person re-identification is still a challenging topic in computer vision because the same person's appearances often undergo significant variance in illumination, camera viewpoint, pose and clothes, while different pedestrians may look very similar due to the similar dressing style. In addition, complex background clutter, occlusion and low resolution also increase the difficulties of person reidentification.

Visualization of HOG features on different color channels. Top row: grayscale map of each channel. Bottom row: HOG model on corresponding channel. The left three channels are form HSV color space, and the right three are from our encoding color space

References

References is not available for this document.