一种多特征联合分布的Camshift目标跟踪算法

A Multi-feature Joint Distribution Camshift Target Tracking Algorithm

针对Camshift算法在纹理与颜色相似干扰、目标遮挡等复杂背景中鲁棒性不高、易出现跟踪丢失等问题,提出了一种多特征联合分布的Camshift目标跟踪算法。新的算法选择等价模式LBP纹理,色调与饱和度为多特征,图像从RGB空间的转化为ULBP-H-S空间。选取图像中运动目标作为目标模板,计算目标模板的ULBP-H联合概率分布图与H-S联合概率分布图,通过自适应系数将两个联合概率分布图按位与运算后到目标的联合概率分布图。在每次迭代搜索中,通过自适应搜索窗口算法预测下一帧的搜索窗口位置与大小,在预测的搜索窗口中使用Camshift算法对目标连续跟踪。实验结果表明,改进的算法能在纹理与颜色相似干扰与目标遮挡复杂环境中,对运动目标跟踪有较高的准确性与鲁棒性。

As the Camshift algorithm has the problem that the tracking object is lost and low robustness in the complex back-grounds such as texture and color similarity interference,target rotation,etc.A multi-feature joint distribution Camshift target track-ing algorithm is proposed.The new algorithm chooses uniform LBP texture,hue and saturation as multi-feature and converts RGB space in the picture to ULBP-H-S space.The moving target in the picture is selected as target template,the ULBP-H joint probabil-ity distribution and H-S joint probability of the target template are calculated,and the target joint probability distribution is calculat-ed by bit-and-operation of two joint probability distribution through adaptive coefficients.In each iterative search process,the adap-tive search window algorithm is used to predict the location and size of search window in the next frame,and the Camshift algorithm is used to continuously track the target in the predicted search window.The experiment results show that the improved algorithm has higher accuracy and robustness in tracking moving target under the complex environment of texture and color similar interference and target occlusion.