ID 原文 译文
14695 利用少量正交导频序列估计出信道初值,通过用户与基站间信道的大尺度衰落系数把用户分簇,根据这些系数按比例地分配接收噪声,再利用数据的统计特性推导出信道衰落参数的均值和方差。 A small number of orthogonal pilot sequences are used to estimate the initial values． Users areclustered into several groups by exploiting the large－scale fading coefficients between users and base station．Thus，the received noise is proportionally assigned according to these coefficients．The mean and variance of channel parameters are then derived using the statistical characteristics of data．
14696 仿真结果表明,当导频数远少于待估计参数的个数时,半盲估计算法的均方误差(Mean Square Error,MSE)优于导频估计的极大似然(Maximum likelihood,ML)算法。 Simulation resultsshow that the mean square error( MSE) of the algorithm is better than that of the maximum likelihood( ML)algorithm when the number of pilots is far less than the estimated parameters．
14697 针对复杂场景下远程视频监控图像异常检测困难、传统算法功能单一(仅针对某种特定场景或某种异常图像进行检测)等问题,提出一种基于深度学习的全自动远程视频异常图像检测方法。 It is difficult to detect abnormal image in complex scene during remote video surveillance and thefunction of traditional method is single( lonly for a specific context or a specific abnormal image) ，a deeplearning based full－automatic method is proposed to detect remote video abnormal images．
14698 首先采用Xavier方法对自行设计的卷积神经网络(Convolutional Neural Network,CNN)的参数进行初始化,然后将标准化后的视频差分图送入CNN的输入层,通过特征提取及下采样,最后在CNN的输出层获得远程视频异常图像检测结果。 Firstly，Xavier isadopted to initialize the parameters of the self－designed convolutional neural network( CNN) ．Then normalized video differential images are sent to the input layer of CNN．Finally，by means of feature extraction anddownsampling，results for abnormal images detection of remote video can be obtained in the output layer ofCNN．
14699 实验结果表明,该方法可以对远程视频监控中突然出现遮挡、模糊和场景切换等多种异常同时进行实时在线检测,准确率可达88.75%。 The experimental results show that the proposed method can conduct real－time online detection of various abnormal images such as image occlusion，blurring and scene switching in the remote video，and theaccuracy rate is up to 88．75% ．
14700 针对在噪声水平比较高的情况下难以从噪声图像本身提取准确先验信息的问题,提出一种从外部干净图像数据集学习非局部自相似先验信息的图像去噪方法。 To solve the problem that it is difficult to extract accurate prior information from the noisy imageitself when the noise level is relatively high，an image denoising method is proposed to learn non－local self－similar prior information from the external clean image dataset．
14701 首先用高斯混合模型学习外部干净图像的非局部自相似先验信息。 Firstly，a Gaussian mixture model is used tolearn the non－local self－similar prior information of the external clean image．
14702 其次利用最大后验概率估计的方法找到与噪声图像块最匹配的外部先验信息,最后利用外部先验对噪声图像块进行稀疏表示。 Secondly，the maximum posterior probability estimation method is used to find the external prior information that best matches the noise image block．
14703 实验对比表明,所提算法在去除噪声的同时可以较好地保留图像的细节信息,使图像数据集的平均峰值信噪比提高0.18 dB以上。 Finally，a sparse representation of noisy image blocks is achieved by external prior． Experimental results show that the proposed algorithm can better retain the details of the image while removing the noise，andimprove the average peak signal－to－noise ratio of the image dataset by more than 0．18 dB．
14704 提出了基于树莓派4B和百度智能云的人脸识别智能门禁系统,该系统由云端服务器、PC机、门禁控制三部分组成。 In the paper, a face recognition intelligent access control system based on raspberry pi 4B and Baidu AI cloud is proposed. The system consists of cloud server, PC and access control.