摘要
提出一种较复杂场景下运动目标跟踪的方法。该方法将Camshift融入到在线Adaboost算法中,特征选取融合纹理轮廓与颜色特征。将在线Adaboost跟踪算法的特征矩阵和分类器运算得到置信图,在置信图上应用Camshift跟踪算法,得到了更加准确的运动目标位置。实验结果表明:在较复杂场景下,对短时间内发生较大形变的、发生遮挡甚至是大面积遮挡的、与背景及其他运动目标颜色相近、快速变化且有加速度的运动目标都能有效跟踪。
A target tracking method in more complex scenarios was presented.The method fuses Camshift into online Adaboost.Feature selection fuses texture contours and color features.Computing the feature matrix of the online Adaboost tracking algorithm with the classifier,the confidence map is getten,then using Camshift on the confidence map,a more accurate position of moving target can be obtained.The experimental results show that,in more complex scenarios,the algorithm is effective for the moving target having large deformation during a short period of time,having blocking and even a large area of blocking,having similar color with the other moving targets or the surroundings,moving fast and having acceleration.
出处
《中南大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2013年第S2期232-238,共7页
Journal of Central South University:Science and Technology
基金
国家高技术研究发展计划("863"计划)项目(2008AA01Z148)
国家自然科学基金资助项目(60975022)
博士点专项科研基金资助(20102304110004)