计算机科学与探索

• 学术研究 •    

强化前景感知的相关滤波目标跟踪

姜文涛,徐晓晴   

  1. 1.辽宁工程技术大学 软件学院,辽宁 葫芦岛125105
    2.辽宁工程技术大学 研究生院,辽宁 葫芦岛 125105

Enhanced foreground perception correlation filtering target tracking

JIANG Wentao, XU Xiaoqing   

  1. 1.School of Software, Liaoning Technical University, Huludao, Liaoning 125105 China
    2.Graduate School, Liaoning Technical University, Huludao, Liaoning 125105 China

摘要: 为了缓解相关滤波跟踪算法在形变、快速运动、运动模糊及相似物干扰等因素影响下导致跟踪精度较低的问题,本文提出了强化前景感知的相关滤波目标跟踪。在相关滤波算法的基础上引入了改进的颜色直方图干扰感知模型。首先,在传统背景对象模型基础上增强前景直方图与背景直方图的颜色差异分量,得到更加突出前景的颜色直方图干扰感知模型。利用相关滤波算法和颜色直方图干扰感知模型分别提取相应特征并计算各自响应,然后通过利用颜色直方图干扰感知模型计算目标区域像素点属于目标的概率均值,控制相关滤波响应和颜色直方图响应的融合权重,最后利用融合后的干扰感知响应图最大值位置定位目标。最后设置跟踪异常判别条件,当异常情况出现,不进行模型更新。当跟踪置信度较高时,则通过帧差法和前后帧间欧氏距离判断目标变化幅度并设置相应的相关滤波模板更新学习率,实现跟踪模板的自适应更新。在OTB100数据集上与主流算法进行实验对比,结果表明本文算法在形变、快速运动、运动模糊及相似物干扰等复杂挑战下相比于其他算法具有更优的跟踪效果及鲁棒性。

关键词: 目标跟踪, 相关滤波, 干扰感知模型, 自适应融合, 模型更新策略

Abstract: In order to alleviate the problem that the tracking accuracy of the correlation filtering target tracking algorithm is low due to the influence of deformation, fast motion, motion blur and similarity interference, this paper proposes the correlation filtering target tracking with enhanced foreground perception. An improved color histogram interference sensing model is introduced based on the correlation filtering algorithm. Firstly, based on the traditional background object model, the color difference component between foreground histogram and background histogram is enhanced to obtain a more prominent foreground color histogram interference sensing model. Using the correlation filter algorithm and color histogram interference perception model and calculate the corresponding feature extracting respective response, and then through the use of color histogram interference perception model calculating the probability of target region pixels belongs to the mean response control correlation filtering and fusion weights of color histogram response, after using the fusion of a maximum of interference response figure perception position location target. Finally, the discriminant conditions of tracking anomalies are set. When abnormal conditions occur, no model update is carried out. When the tracking confidence is high, the range of target change is judged by frame difference method and Euclidean distance between two frames and the corresponding learning rate of correlation filtering template is set to realize the adaptive updating of tracking template. Experimental results on OTB100 data set show that the proposed algorithm has better tracking performance and robustness than other algorithms under complex challenges such as deformation, fast motion, motion ambiguity and similarity interference.

Key words: Target tracking, Correlation filter, Interference perception model, Adaptive fusion, Model update strategy