ID 原文 译文
17485 然后在考虑阵元误差的情况下，基于最小二乘准则迭代地估计杂波表示系数和阵元误差， Next,with the consideration of array error, it estimates iteratively the clutter representation coefficient and arrayerror by the least square criterion.
17486 最后利用估计得到的最优杂波表示系数和阵元误差直接在阵元脉冲域进行杂波对消。 Finally, the clutter cancellation is conducted by the obtained optimal clutterrepresentation coefficient and array error in element-pulse domain.
17487 该方法无须估计待检测单元统计特性；没有孔径损失； The proposed method does not need to estimate the statistical properties of cell under test and has no aperture loss.
17488 不需要训练样本；即使在距离模糊情况下也能有效地抑制密集目标环境下机载面阵雷达回波数据中的非均匀杂波。 In addition, it does not need anytraining sample and can suppress effectively the heterogeneous clutter of airborne planar array radar echo datain rich target environment even if range ambiguity exists.
17489 仿真结果验证了该文方法的有效性。 Simulation results verify the validity of the proposed method.
17490 近年来，采用孪生网络提取深度特征的方法由于其较好的跟踪精度和速度，成为目标跟踪领域的研究热点之一， In recent years, the method of extracting depth features from siamese networks has become one of thehotspots in visual tracking because of its balanced in accuracy and speed.
17491 但传统的孪生网络并未提取目标较深层特征来保持泛化性能，并且大多数孪生网络只提取局部领域特征，这使得模型对于外观变化是非鲁棒和局部的。 However, the traditional siamesenetwork does not extract the deeper features of the target to maintain generalization performance, and mostsiamese architecture networks usually process one local neighborhood at a time, which makes the appearancemodel local and non-robust to appearance changes.
17492 针对此，该文提出一种引入全局上下文特征模块的DenseNet孪生网络目标跟踪算法。 In view of this problem, a densenet-siamese network withglobal context feature module for object tracking algorithm is proposed.
17493 该文创新性地将DenseNet网络作为孪生网络骨干，采用一种新的密集型特征重用连接网络设计方案，在构建更深层网络的同时减少了层之间的参数量，提高了算法的性能， This paper innovatively takes densenetnetwork as the backbone of siamese network, adopts a new design scheme of dense feature reuse connectionnetwork, which reduces the parameters between layers while constructing deeper network, and enhances thegeneralization performance of the algorithm.
17494 此外，为应对目标跟踪过程中的外观变化，该文将全局上下文特征模块(GC-Model)嵌入孪生网络分支，提升算法跟踪精度。 In addition, in order to cope with the appearance changes in theprocess of object tracking, the Global Context feature Module (GC-Model) is embedded in the siamese networkbranches to improve the tracking accuracy.