中文摘要：

针对目标跟踪过程中发生遮挡时跟踪出现偏差的情况，在传统MeanShift跟踪算法的基础上引入极限学习机（extreme learning machine，ELM），提出了基于ELM与Mean Shift的目标跟踪算法。该算法根据过去3个时刻的目标位置信息。利用ELM预测出目标当前可能位置，代替目标前一时刻位置作为Mean Shift迭代起始点，并在邻域范围内进行迭代，得到目标的真实位置。实验结果表明，与现有的改进算法相比，新算法减少了迭代次数和运算时间，同时在遮挡情况下能够准确定位目标位置并进行跟踪，提高了跟踪系统的实时性和鲁棒性。

英文摘要：

According to the characteristics of the load of the lateral embankment filling on the culvert,based on the limit equilibrium theory and the ultimate bearing capacity formula of Meyehof and Hanna,the formula for calculating ultimate bearing capacity of the culvert foundation with crust layer was deduced. A series of comparisons were carried out among the calculation results obtained from Hansen＇s weighted average method,the stress diffusion angle method,the improved formula proposed in this paper and the finite element method.And the influences of the height of embankment filling,the thickness of crust layer,the ratio of cohesive and the internal friction angle of the crust layer to the subjacent soil layer on the ultimate bearing capacity were investigated,in addition,a comparison of the results was made from engineering practical examples and the calculation results from the four theoretical calculation methods. The results show that with the increase of height of embankment filling and the thickness of crust layer,the results of the foundation bearing capacity calculated by the proposed improved formula method and finite element method are the closest. The influence of the variations of cohesive ratio and internal friction angle ratio on the calculated results of the proposed formula method is more consistent with that of the finite element method. The results of the proposed formula method are more approximate to the engineering practical examples than Hansen＇s weighted method and the stress diffusion angle method.