ID 原文 译文
6144 最后对符号周期估计算法进行性能比较。仿真结果表明,在低信噪比下,两种算法均能有效实现FBMC-OQAM信号的符号周期估计。 Finally, the symbol period estimation algorithm for performance comparison.The simulation results show that under low SNR, the two algorithms can effectively symbols FBMC - OQAM signal cycle is estimated.
6145 针对传统雷达图像目标检测方法在海杂波及多种干扰物组成的复杂背景下目标分类识别率低、虚警率高的问题,提出将当前热点研究的深度学习方法引入到雷达图像目标检测。 In view of the traditional radar image target detection in sea clutter or multiple ways of distractors targets under the complex background of classification recognition rate is low, the problem of high false alarm rate, the depth of the current hot research learning method is introduced into the image target detection radar.
6146 首先分析了目前先进的YOLOv3检测算法优点及应用到雷达图像领域的局限,并构建了海杂波环境下有干扰物的舰船目标检测数据集。 First analysis of the current advantages and advanced YOLOv3 detection algorithm is applied to the limitations in the field of radar image, and build the sea clutter environment of distractors ship target detection data set.
6147 数据集包含了不同背景、分辨率、目标物位置关系等条件,能够较完备地满足实际任务需要。 Data set contains a different background, resolution, the relationship between the target position condition, able to meet the needs of the actual task is complete.
6148 针对该数据集包含目标稀疏、目标尺寸小的特点,首先利用K-means算法计算适合该数据集的锚点坐标; For the data set contains sparse of target, the target characteristics of small size, first using the K - means algorithm is suitable for the data set the anchor point coordinates;
6149 其次在YOLOv3的基础上提出改进多尺度特征融合预测算法,融合了多层特征信息并加入空间金字塔池化。 Secondly in YOLOv3 improved multi-scale feature fusion is proposed on the basis of prediction algorithm, combined with multi-level pyramid pooling feature information and join space.
6150 通过大量对比实验,在该数据集上,所提方法相比原YOLOv3检测精度提高了6.07%。 Through a large number of experiments, on the data set, the proposed method compared with the original YOLOv3 precision increased by 6.07%.
6151 面向双层无线传感器网络覆盖质量评估,设计出基于模糊小波聚类混合的多目标覆盖质量评估方法。 Facing the double wireless sensor network coverage quality evaluation, designed based on fuzzy wavelet hybrid multi-objective covering clustering quality evaluation method.
6152 建立网络单元概念和双层网络模型,在各汇聚节点开展各子目标预处理。 Concept and double unit network model is set up in each node in each subgoal pretreatment.
6153 集中建立二次预警机制:设计基于模糊小波神经网络的分析融合子系统,实现一次预警,选出显著低效覆盖单元; Concentrated establish an early warning system for the secondary: design based on fuzzy wavelet neural network analysis subsystem, realize a warning, select significant inefficiencies cover unit;