摘要
针对当前行人再识别在度量学习算法中使用的主成分分析法容易丢失分类信息的问题,提出一种基于增量式线性判别分析的行人再识别算法.算法采用线性判别分析的映射方法使样本在投影子空间中能够保持最大化的分类信息,并利用增量学习的方法使度量学习模型能够根据新标记的训练样本进行更新.方法不仅考虑了映射子空间保留样本分类信息的问题,而且考虑了度量矩阵对新样本的更新性.仿真结果表明,该方法不仅能增强算法的准确性,具有较高的行人再识别率,而且对新样本还具有可扩展性.
Aiming for person re-identification using principal component analysis in metric learning which easily causes loss of classification information,a new person redentification algorithm is proposed based on incremental linear discriminant analysis. The proposed method introduce a mapping method of linear discriminant analysis to maintain the maximize classification information in projection subspace. Meanwhile, it introduces metric learning method to enable the metric learning model updating according to the training set of the new training samples. The proposed method not only considers retaining sample classification information in projection subspace, but also considers the update ability of the metric model, which can enhance the accuracy and extensibility of the algorithm. Compared with the similar algorithm,the method can reach a high person redentification rate.
出处
《小型微型计算机系统》
CSCD
北大核心
2017年第3期595-600,共6页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61104213
61573168)资助
江苏省产学研前瞻性联合研究项目(BY2015019-15)资助
关键词
线性判别分析
增量学习
KISSME算法
行人再识别
linear discriminant analysis
incremental learning
KISSME algorithm
person re-identification
