基于SURF特征和多示例学习的目标跟踪算法
首发时间:2016-06-20
摘要:视频目标跟踪是机器视觉与人工智能的重要研究方向。本文提出一种基于SURF特征和多示例学习(Multiple Instance Learning,MIL)的目标跟踪算法。首先提取感兴趣目标及其周围图像的SURF特征;然后将SURF描述子引入到MIL中进行潜在语义分析LSA(Latent Semantic Analysis)获得正负包的潜在语义特征;最后,利用支撑向量机对所获得的潜在语义特征训练分类器,将跟踪中MIL问题转化在线学习,进而判断是否为跟踪目标。实验结果表明本文算法对于目标存在尺度、光照、姿态等因素变化时依然能获得较好的跟踪效果。
关键词: 目标跟踪 多示例学习 潜在语义分析; SURF特征
For information in English, please click here
Object tracking based on multi-instance learning with SURF features
Abstract:Object tracking is an important research in machine vision and artificial intelligence. In this paper a target tracking algorithm based on the SURF and MIL is proposed. Firstly, we extract the SURF features of the target of interest and its surrounding image; secondly, SURF descriptor will be introduced into the MIL for latent semantic analysis, and the latent semantic features of the positive and negative bags will be obtained; thirdly, the two kinds of latent semantic features then are input into SVM, by which the MIL in tracking would be transited to online learning; the object of interest will be tracked finally. The experimental results shows the robustness and efficiency of the proposed algorithm under the situation of the variation of the scale, luminance and gesture.
Keywords: Object tracking Multiple instance learning Latent sematic analysis SURF
论文图表:
引用
No.4696839112714114****
同行评议
共计0人参与
勘误表
基于SURF特征和多示例学习的目标跟踪算法
评论
全部评论0/1000