ID 原文 译文
11724 在基于直方图的序列图像目标跟踪算法中,目标的直方图通常都是在跟踪初始化时从目标所在的区域获得,然而单个直方图难以适应跟踪全过程中目标的各种变化。 In the sequence image target tracking algorithm based on histogram, target histogram is usually in tracking initialization time obtained from the target area, however, a single histogram is difficult to adapt to track the whole process of all kinds of changes in the target.
11725 针对事先已知目标几种典型外观的跟踪问题,提出了一种基于粒子滤波器的多直方图尺度空间跟踪算法。 For several typical appearance of known to target tracking problem, this paper proposes a multiple histogram scale space based on particle filter tracking algorithm.
11726 利用多个典型直方图的线性加权来表示目标的直方图,根据目标的当前区域估计加权系数,生成下一帧的目标概率分布图。 Using several typical histogram linear weighted histogram to represent the goals, estimate the weighted coefficient according to the target of the current area, to generate the next frame target probability distribution.
11727 在目标概率分布图上运用尺度空间粒子滤波器,来估计多尺度规范化Laplacian滤波函数的极值,从而实现目标的定位。 Using scale space on target probability distribution of the particle filter, for estimation of multi-scale normalized Laplacian filtering function extreme value, thus achieve the goal of positioning.
11728 通过在真实序列上与现有算法的对比,表明了此算法不仅可以适应目标的色彩和明暗变化,而且能更准确地描述目标的大小,显著提高跟踪的精度。 By contrast with existing algorithms on real sequence, shows that this algorithm can not only adapt to the target of colour and light and shade change, and can be more accurate to describe the size of the object, significantly improve the tracking precision.
11729 针对小子样产品加速因子难以确定问题,提出了一种基于寿命分布和贝叶斯的加速因子确定方法。 In view of the small sample product accelerated factor is difficult to determine, is put forward based on the life distribution and bayesian method of determining the acceleration factor.
11730 首先根据相似产品内场贮存数据信息,对正常应力下寿命分布参数进行极大似然估计(maximum likelihood estimation,MLE)。 ‭First according to the similar product diamond store data information, life under the normal stress distribution parameters on maximum likelihood estimation (maximum likelihood estimation, the MLE).
11731 再利用Fisher信息矩阵求逆得到其方差-协方差矩阵,最终得到正常应力下寿命分布参数的分布。 Reuse of Fisher information matrix inversion to get the variance - covariance matrix, finally get the normal stress distribution of life distribution parameters.
11732 然后根据新研产品加速应力下小子样试验信息结合相似产品加速应力试验信息和专家经验等先验信息,利用ML-Ⅱ方法确定混合先验分布的权重。 And the then accelerated stress under small sample test, according to new research product information combined with similar product accelerated stress test information and expert experience a priori information, such as ML - Ⅱ method is used to determine the mixed weight of the prior distribution.
11733 再利用Bayes方法得到产品寿命分布参数的后验分布。 Recycling the Bayes method to get product life distribution parameters of the posterior distribution.