Computer Engineering and Applications ›› 2019, Vol. 55 ›› Issue (18): 146-150.DOI: 10.3778/j.issn.1002-8331.1808-0203

Previous Articles     Next Articles

Multi-Feature Subspace Person Re-Identification Based on BOW Model

ZHU Xiaobo, CHE Jin   

  1. 1.School of Physics and Electronic-Electrical Engineeing, Ningxia University, Yinchuan 750021, China
    2.Ningxia Key Laboratory of Intelligent Sensing for Desert Information(Ningxia University), Yinchuan 750021, China
  • Online:2019-09-15 Published:2019-09-11

融合BOW模型的多特征子空间行人重识别算法

朱小波,车进   

  1. 1.宁夏大学 物理与电子电气工程学院,银川 750021
    2.宁夏沙漠信息智能感知重点实验室(宁夏大学),银川 750021

Abstract: Aiming at the problems of current person re-identification in the target appearance characteristics and measurement algorithm, a multi-feature subspace person re-identification based on BOW model is proposed. The 2-D Gaussian template is used to weaken the image background on the person image. The BOW feature descriptor and the YUV+HSV color feature descriptor are extracted and fused to form the final feature descriptor. In the similarity measure, the similarity measure is performed by learning a subspace in the original feature space and learning the measure matrix in the subspace. The experimental results on the two datasets of VIPeR and CUHK01 show that the proposed algorithm can significantly improve the person re-identification rate.

Key words: BOW model, subspace, weaken the image background, person re-identification

摘要: 针对目前行人重识别算法在目标外观特征和度量算法方面的问题,提出一种融合BOW模型的多特征子空间行人重识别算法。在行人图像上采用2-D高斯模板将图像背景弱化,然后提取BOW特征描述子和YUV+HSV颜色特征描述子,并将其融合组成最终的特征描述子。在相似性度量方面,采用在原始特征空间学习一个子空间,并在该子空间学习测度矩阵的方法进行相似性度量。在VIPeR和CUHK01两个数据集上的实验结果表明,提出的算法能够明显地提高行人重识别率。

关键词: BOW模型, 子空间, 背景弱化, 行人重识别