ID	原文	译文
1423	半监督项约束了有标记像素的类标在分类过程中保持不变。	The semi-supervision term enforces that class information of labeled pixels keep fixed during the classification.
1424	类标光滑项约束了像素间类标的光滑性。	The pairwise smoothness term enforces class label smoothness.
1425	基于对 PauliRGB 图像进行超像素分割而产生的初始化类标图，交替迭代优化所定义的能量函数直至其收敛。	Started from an initialized class label map generated using superpixel seg-mentation of polarimetric PauliRGB image, the proposed method iteratively optimizes the energy function until the modelconverges.
1426	在两幅真实极化 SAR 图像上的实验结果表明，该算法达到了优异的分类效果，其性能优于当前已有算法。	Experimental results conducted on two benchmark PolSAR images show that our approach effectively improvesthe classification accuracy with limited numbers of labeled pixels.
1427	智能环境传统的规则推理机制中，网关内布置的推理机从各种传感器中获取推理所需数据并与规则库相匹配，承担整个推理工作。	Under traditional rule-based reasoning in smart environment, the inference engine deployed in gateway col-lects data from various sensors to match the rules, which undertakes whole reasoning work.
1428	本文利用 Rete 算法将规则构建为推理网络，并结合雾计算的概念将 Rete 推理节点分配至环境内配置的智能节点中协同推理以减轻网关负载，由此推理节点的分配成为关键，分配不合理将导致资源利用不平衡及响应延迟。	In current research, we use Retealgorithm to construct an inference network by rules and then allocate the Rete inference nodes to smart nodes for collabora-tive reasoning based on fog computing, therefore, the allocation mechanism becomes crucial.
1429	本文利用活动影响下规则触发的规律设计了活动聚类算法 CoA(Clustering of Activities)对活动聚类后分别建立其推理网络，计算出智能节点之间的最短路径后将结果代入针对其层次延迟性而设计的分配算法 AAoRN(Allocation Algorithm of Rete Inference Nodes)，从而将推理节点最优分配至各个智能节点。	In this paper, we utilize the reg-ularity of rules triggered under the influence of activities to design an algorithm CoA(Clustering of Activity), which clustersactivities and respectively constructs the inference networks, subsequently, we calculate the shortest path between smart nodesand substitute the results into AAoRN(Allocation Algorithm of Rete Inference Nodes), which is proposed to overcome hier-archical delay for optimally allocating each inference node.
1430	理论分析和实验结果表明，本文机制在有效利用智能节点资源的同时降低了大致 55% 的延迟。	Theoretical analysis and experimental results show that the pro-posed mechanism efficiently utilizes the resources and has reduced the delay by about 55% .
1431	为了实现自动确定类别数的高精度遥感影像分割，提出一种自适应类别的层次高斯混合模型(Hierar-chical Gaussian Mixture Model，HGMM)遥感影像分割算法。	To accurately segment remote sensing image and automatically determine the number of classes, a Hierar-chical Gaussian Mixture Model (HGMM)based remote sensing image segmentation algorithm with self-adaptive number ofclasses is proposed.
1432	提出算法采用多个高斯混合模型加权和定义 HGMM，用于建模具有非对称，重尾和多峰等复杂特性的影像统计模型。	In the proposed algorithm, HGMM is defined by the weighted sum of several Gaussian Mixture Models(GMM), which is used to model the asymmetric, heavy-tailed and multimodal distributions of image.