| Related citation: | Zhongli Wang,Chunxiao Jia,Baigen Cai,Litong Fan,Chuanqi Tao,Zhiyi Zhang,inling Wang,Min Zhang.A Novel Tracking-by-Detection Method with Local Binary Pattern and Kalman Filter[J].Journal of Harbin Institute Of Technology(New Series),2018,25(3):74-87.DOI:10.11916/j.issn.1005-9113.16185. |
|
| Author Name | Affiliation | | Zhongli Wang | School of Electronics and Information Engineering, Beijing Jiaotong University, Beijing 100044, China | | Chunxiao Jia | School of Electronics and Information Engineering, Beijing Jiaotong University, Beijing 100044, China | | Baigen Cai | School of Electronics and Information Engineering, Beijing Jiaotong University, Beijing 100044, China | | Litong Fan | School of Electronics and Information Engineering, Beijing Jiaotong University, Beijing 100044, China | | Chuanqi Tao | CSR Qindao Sifang Co., Ltd., Qindao 266111, China | | Zhiyi Zhang | CSR Qindao Sifang Co., Ltd., Qindao 266111, China | | inling Wang | CSR Qindao Sifang Co., Ltd., Qindao 266111, China | | Min Zhang | CSR Qindao Sifang Co., Ltd., Qindao 266111, China |
|
| Abstract: |
| Tracking-Learning-Detection (TLD) is an adaptive tracking algorithm, which tracks by learning the appearance of the object as the video progresses and shows a good performance in long-term tracking task. But our experiments show that under some scenarios, such as non-uniform illumination changing, serious occlusion, or motion-blurred, it may fails to track the object. In this paper, to surmount some of these shortages, especially for the non-uniform illumination changing, and give full play to the performance of the tracking-learning-detection framework, we integrate the local binary pattern (LBP) with the cascade classifiers, and define a new classifier named ULBP (Uniform Local Binary Pattern) classifiers. When the object appearance has rich texture features, the ULBP classifier will work instead of the nearest neighbor classifier in TLD algorithm, and a recognition module is designed to choose the suitable classifier between the original nearest neighbor (NN) classifier and the ULBP classifier. To further decrease the computing load of the proposed tracking approach, Kalman filter is applied to predict the searching range of the tracking object. A comprehensive study has been conducted to confirm the effectiveness of the proposed algorithm (TLD_ULBP), and different multi-property datasets were used. The quantitative evaluations show a significant improvement over the original TLD, especially in various lighting case. |
| Key words: Tracking-Learning-Detection (TLD) local binary pattern (LBP) Kalman filter |
| DOI：10.11916/j.issn.1005-9113.16185 |
| Clc Number:TP391.4 |
| Fund: |
|
| Descriptions in Chinese: |
| 一种基于LBP和KF增强的检测-跟踪方法 #$TAB王忠立1，贾春肖1，#$TAB蔡伯根1，樊俐彤1，陶传奇2，张志毅2，王银灵2，张敏2,吕国艳2 (1．北京交通大学，电子信息工程学院，北京 100044； 2．青岛四方车辆股份有限公司，青岛 266111) 创新点说明：1）针对TLD算法存在对光照变换时稳定性不够好的问题，通过试验分析发现，原算法中检测器的性能对此有很大影响。本文将LBP特征描述子集成到算法的Tracking-by-detection框架中，通过一个判别模块来确定是采用NN检测器，还是LBP检测器，较好的解决了原算法的这一问题。实验证明，改进后的算法，正确性、召回率等指标都有较大改善。 2）TLD算法的计算复杂度较高，不能实时跟踪。经过LBP特征描述子改进后的算法（TLD_ULBP）复杂度也略高于原来的算法。为了提高算法效率，采用Kalman滤波来预测目标搜索区域。论文给出了KF滤波预测的详细过程及策略，并对算法的性能进行了评估。 3）通过大量实验对改进后的算法在准确率、召回率、F-measure等指标上进行了验证，除极个别测试视频外，绝大多数视频下都有较大改善。另外，通过实验发现，经过KF加速，所提出的方法在实时性和精度上都有较好提升。 关键词：TLD算法；LBP；卡尔曼滤波 |
