ID 原文 译文
50017 并通过增量式强化集成学习，获得最后的场景分类。 and through the incremental strengthen integration study, obtain the final scene classification.
50018 在两个标准的高分辨率遥感图像数据集上的实验结果表明，MNCC算法具备较好场景分类结果。 In two standards of high-resolution remote sensing image data set on the experimental results show that the algorithm of MNCC have a good scene classification result.
50019 针对基于单一颜色特征的粒子滤波跟踪算法易受光照变化、部分遮挡及相似干扰物的影响，而利用多特征融合的粒子滤波方法存在各特征权值、跟踪模板及窗口大小自适应选取问题，提出了一种基于模糊测度的多特征融合鲁棒粒子滤波跟踪算法。 Particle filter tracking algorithm based on single color features are susceptible to illumination changes, partial sheltering and similarity, the effect of distractors, and the use of multiple features fusion particle filter method has the feature weights, tracking template and adaptive window size selection problem, this paper proposes a more robust feature fusion based on fuzzy measure particle filter tracking algorithm.
50020 采用颜色及边缘方向直方图来描述目标量测模型，通过分别计算这两类特征在候选目标与参考目标之间的Bhattacharyya距离来确定其各自特征的模糊测度，通过查取模糊规则表来自适应地确定两类特征的权重; Color and edge direction histogram is used to describe the target measurement model, through the two types of features are calculated respectively in the candidate Bhattacharyya distance between the target and reference target to determine the characteristics of each fuzzy measure, fuzzy rules by pick up table from adapt to determine the weights of the two kinds of characteristics;
50021 将连续帧的多特征联合模板更新机制用于对初始目标模板的更新; Updating mechanism will be more characteristics of continuous frame joint is used to update on the initial target template;
50022 针对目标发生尺度变化造成跟踪窗口难以自适应的问题，通过引入粒子离散度实现了跟踪窗尺寸的自适应调整。 Scale changes in target to cause the problem of adaptive tracking window, by introducing the particle discrete degree to realize adaptive adjustment of the tracking window size.
50023 实验结果表明:所提出的跟踪算法位置平均误差小于8个像素，相比于传统方法可以有效克服光照、部分遮挡以及相似目标干扰等影响，具有较高的跟踪精度及较强的鲁棒性。 The experimental results show that the proposed tracking algorithm position average error is less than eight pixels, compared with the traditional method can effectively overcome the interference such as illumination, partial sheltering and similar objectives, has higher tracking precision and strong robustness.
50024 针对运动目标跟踪存在的目标遮挡和光照变化问题，提出一种基于压缩感知的粒子滤波跟踪算法。 Aiming at the existence of movement target tracking in the object shelter and illumination change problem, put forward a kind of particle filter tracking algorithm based on compression perception.
50025 将改进的压缩感知跟踪算法提取的特征融合到粒子滤波跟踪框架中，并对压缩感知提取的特征和原始粒子滤波中的颜色特征进行可信度判定，能够较好地处理图像序列中由于目标遮挡和光照变化所带来的影响。 Compressed awareness will improve tracking algorithm to extract the characteristics of the merged into the field of particle filter tracking framework, and the compression perception to extract features and color features of the original particle filter for reliability discrimination, can well deal with because of target in image sequence block, and the effects of illumination change.
50026 此算法在公开数据库中进行测试，实验结果表明，提出的算法与已有改进压缩感知跟踪算法和粒子滤波跟踪算法相比，鲁棒性更好，能准确实时地对目标进行跟踪。 This algorithm in the public database, the experimental results show that the proposed algorithm and improved compression perception compared tracking algorithm and the particle filter tracking algorithm, better robustness and can accurately in real time tracking of target.