ID 原文 译文
14865 经实验，该方法在无线传感器网络能量均衡度等方面应用价值比较高。 The experimental results show that this method has high application valuein energy balance degree of wireless sensor network.
14866 针对如何有效地从细胞形态图中提取特征指标，与其他细胞形态图形区分开来，判别细胞是否产生病变。 How to effectively extract the characteristic index from the cell morphology graph, anddistinguish it from other cell morphology graphs to judge whether the cell has pathological changes.The cell image is preprocessed by chroma interval method, smoothing, binarization, hole filling anddenoising.
14867 细胞图像经过色度区间法、平滑处理、二值化、孔洞填充、去噪声的预处理，再对细胞图像的区域标记，进行形状因子、宽长比、矩形度、紧凑度、重心位置的形态学计算，BP 神经网络作为识别算法，通过 Visual Studio 程序的处理，提供统计数据，作为病理诊断的一项重要依据。 Then, the region of the cell image is marked, and the morphological calculation of shapefactor, aspect ratio, rectangularity, compactness and barycenter position is carried out. BP neuralnetwork is used as the recognition algorithm, and Visual Basic is used as the recognition tool Studioprogram provides statistical data as an important basis for pathological diagnosis.
14868 管道焊缝质量是焊接工程质量评定中的重要一项，因此焊缝质量检测在焊接质量评定中具有重要的地位。 Pipeline weld quality is an important item in welding engineering quality assessment,so weld quality inspection has an important position in welding quality assessment.
14869 为了有效追踪焊缝质量，提出一种改进的单尺度 Retinex 算法，对图像照度分量进行空域和频域低通滤波，并采用非线性变换函数调整反射分量的数据范围，来消除光照不匀对图像质量造成的影响。 In order toeffectively track the weld quality, an improved single-scale Retinex algorithm is proposed, whichperforms low-pass filtering of the image illuminance components in the spatial and frequency domains,and uses a nonlinear transformation function to adjust the data range of the reflected componentsto eliminate uneven illumination The impact of quality.
14870 仿真实验结果表明，改进的单尺度 Retinex 算法处理效果优于 SSR 和多尺度 Retinex 算法（MSR），处理后的管道焊缝图像能够对焊缝内表面质量的判定提供很好的依据。 The simulation experiment results show thatthe improved single-scale Retinex algorithm is better than SSR and multi-scale Retinex (MSR). Theprocessed pipeline weld image can provide a good basis for the judgment of the internal surfacequality of the weld.
14871 为实现简单手势的识别，采用机器学习的方式。 In order to realize the recognition of simple gestures, the machine learning method isadopted.
14872 在通过摄像头获取图片后，运用了 SURF 局部特征提取算法来提取图像的特征点，接着通过 K-means 聚类形成视觉词汇构建 BoW 模型，接着通过 SVM 进行分类识别，识别率较高。 After the images were acquired by the camera, SURF local feature extraction algorithm wasused to extract the feature points of the image. Then k-means clustering was used to form a visualvocabulary to construct the BoW model, and then SVM was used for classification and recognition, witha high recognition rate.
14873 本文研究了一种基于双层搜索空间聚类（Two-Level Search Space Clustering, TL-SSC）的移动声源定位与跟踪方法，采用 TL-SSC 方法和平均值计算方法来确定移动声源在离散时间域上的坐标，并且在声源短暂停止发声时能够通过速度与加速度计算进行位置跟踪。 This paper introduced a moving sound source localization and tracking method based on twolevel search space clustering (TL-SSC). The average calculation method is used to determine thecoordinates of moving sound source in discrete time domain, and the position can be tracked bycalculating the velocity and acceleration when the sound source stops sounding briefly.
14874 仿真实验结果证实了该方法在不同混响环境下能够保持定位的准确性，满足移动声源实时定位的需求。 Simulationresults show that this method can maintain the accuracy of positioning in different reverberationenvironments and meet the needs of real-time localization of moving sound sources.