ID 原文 译文
19215 实验测试表明：跳频电台的暂态信号时间序列属于混沌时间序列，同时实测多跳频信号的电台分选结果证明了Low-rank聚类算法在跳频电台分选上的可行性。 The experimental tests show that the transient signal of the frequency-hopping transmitter belongsto the chaotic time series. At the same time, the classification results of the frequency-hopping signals demonstrate the feasibility of the Low-rank clustering algorithm in frequency-hopping transmitter classification.
19216 基于卷积神经网络的图像篡改检测算法利用卷积神经网络的学习能力可以实现不依赖于单一图像属性的图像篡改检测，弥补传统图像篡改检测方法依赖单一图像属性、适用度不高的缺陷。 The image forgery detection algorithm based on convolutional neural network can implement theimage forgery detection that does not depend on a single image attribute by using the learning ability ofconvolutional neural network, and make up for the defect that the previous image forgery detection algorithmrelies on a single image attribute and has low applicability.
19217 利用深层多神经元的单一网络结构的图像篡改检测算法虽然可以学习更高级的语义信息，但检测定位篡改区域效果并不理想。 Although the image forgery detection algorithmusing a single network structure of deep layers and multiple neurons can learn more advanced semanticinformation, the result of detecting and locating forgery regions is not ideal.
19218 该文提出一种基于级联卷积神经网络的图像篡改检测算法，在卷积神经网络所展示出来的普遍特性的基础上进一步探究其深层次的特性，利用浅层稀神经元的级联网络结构弥补以往深层多神经元的单一网络结构在图像篡改检测中的缺陷。 In this paper, an image forgerydetection algorithm based on cascaded convolutional neural network is proposed. Based on the generalcharacteristics exhibited by convolutional neural network, and then the deeper characteristics are furtherexplored. The cascaded network structure of shallow layers and thin neurons figures out the defect of the singlenetwork structure of deep layers and multiple neurons in image forgery detection.
19219 该文提出的检测算法由级联卷积神经网络和自适应筛选后处理两部分组成，级联卷积神经网络实现分级式的篡改区域定位，自适应筛选后处理对级联卷积神经网络的检测结果进行优化。 The proposed detectionalgorithm in this paper consists of two parts: the cascade convolutional neural network and the adaptivefiltering post-processing. The cascaded convolutional neural network realizes hierarchical forgery regions localization, and then the adaptive filtering post-processing further optimizes the detection result of the cascaded convolutional neural network.
19220 通过实验对比，该文算法展示了较好的检测效果，且具有较高的鲁棒性。 Through experimental comparison, the proposed detection algorithmshows better detection results and has higher robustness.
19221 雷达目标的后向散射对目标姿态与雷达视线的相对几何关系十分敏感，同一目标相对于雷达视线的姿态不同时，散射特性十分不同。 The backscattering of the radar targets is sensitive to the relative geometry between orientations ofthe targets and the radar line of sight. When the orientations of the same target are different from the radarline of sight, the scattering characteristics are quite different.
19222 倾斜地表和倾斜建筑物等目标可能扭转后向散射回波的极化基，进而导致交叉极化分量过高，图像体散射成分过估计。 Targets such as inclined ground and inclinedbuildings may reverse the polarization base of the backscattered echo, which causes the cross-polarizationcomponent to be too high and the volume scattering component of the image is overestimated.
19223 该文针对图像体散射成分过估计的现象，提出一种基于极化特征参数()和极化干涉相似性参数(PISP)的极化干涉分解方法。 In this paper, apolarimetric interferometric decomposition method based on polarimetric parameters () and PolarimetricInterferometric Similarity Parameters (PISP) is proposed to solve the overestimation problem.
19224 该方法充分考虑了散射体在雷达视线方向上的散射多样性，对不同取向的倾斜地表和倾斜建筑物等目标产生的交叉极化分量进行更好的适配，得到更好的分解结果。 The method makes full use of the scattering diversity of the scatterer in the radar line of sight. The cross-polarization components generated by targets such as inclined grounds and inclined buildings with different orientations are better adapted to obtain better decomposition results.