A new conflicting evidence fusion method is proposed for the deficiency of Dempster's rule which can not fuse the conflicting evidence. Evidence is divided into three categories:believable evidence, non-conflictin...A new conflicting evidence fusion method is proposed for the deficiency of Dempster's rule which can not fuse the conflicting evidence. Evidence is divided into three categories:believable evidence, non-conflicting evidence and conflicting evidence. The influences of these three categories of evidences on fusion results when discounted are analyzed respectively. On these bases, the evidence distance and the conjunctive conflict are utilized in sequence to recognize the believable evidence and non-conflicting evidence. The discounting factors of these two categories of evidences are set to one, which keeps the evidences support the true hypothesis to the greatest degree, and makes the fusion results focus onto the true hypothesis. Examples of some missile fault diagnosis show that the new method can effectively fuse the conflicting evidences, and is suited to fuse the relievable evidences. The new method improves the reliability and rationality of fusion results compared with traditional methods.展开更多
针对视频情感识别中存在运算复杂度高的缺点,提出一种基于时空局部二值模式矩(Temporal-Spatial Local Binary Pattern Moment,TSLBPM)的双模态情感识别方法。首先对视频进行预处理获得表情和姿态序列;然后对表情和姿态序列分别提取TSL...针对视频情感识别中存在运算复杂度高的缺点,提出一种基于时空局部二值模式矩(Temporal-Spatial Local Binary Pattern Moment,TSLBPM)的双模态情感识别方法。首先对视频进行预处理获得表情和姿态序列;然后对表情和姿态序列分别提取TSLBPM特征,计算测试序列与已标记的情感训练集特征间的最小欧氏距离,并将其作为独立证据来构造基本概率分配(Basic Probability Assignment,BPA);最后使用Dempster-Shafer证据理论联合规则得到情感识别结果。在双模态表情和姿态情感数据库上的实验结果表明,本文提出的时空局部二值模式矩可以快速提取视频图像的时空特征,能有效识别情感状态。与其他方法的对比实验也验证了本文融合方法的优越性。展开更多
基于弹着点空间分布对目标毁伤效能的差异化影响,构建导弹命中目标不同重要区域的概率分布模型,实现对传统命中精度概念的扩展。针对导弹实打试验过程复杂、费用高、次数少的实际,采用贝叶斯方法融合多源信息,基于区域划分—分布确定—...基于弹着点空间分布对目标毁伤效能的差异化影响,构建导弹命中目标不同重要区域的概率分布模型,实现对传统命中精度概念的扩展。针对导弹实打试验过程复杂、费用高、次数少的实际,采用贝叶斯方法融合多源信息,基于区域划分—分布确定—先验融合—后验求解的思路进行导弹命中精度估计。选取Dirichlet分布作为命中精度参数的先验分布,运用D-S(Dempster-Shafer)证据理论对先验信息进行融合处理,基于马尔可夫链蒙特卡罗(Markov chain Monte Carlo,MCMC)方法对精度参数的后验分布进行求解。示例表明,该方法能够细致描述导弹命中目标不同重要区域的概率,并科学融合多源命中精度先验信息,为导弹命论文拓展中精度估计方法及测试方案优化提供理论借鉴。展开更多
文摘A new conflicting evidence fusion method is proposed for the deficiency of Dempster's rule which can not fuse the conflicting evidence. Evidence is divided into three categories:believable evidence, non-conflicting evidence and conflicting evidence. The influences of these three categories of evidences on fusion results when discounted are analyzed respectively. On these bases, the evidence distance and the conjunctive conflict are utilized in sequence to recognize the believable evidence and non-conflicting evidence. The discounting factors of these two categories of evidences are set to one, which keeps the evidences support the true hypothesis to the greatest degree, and makes the fusion results focus onto the true hypothesis. Examples of some missile fault diagnosis show that the new method can effectively fuse the conflicting evidences, and is suited to fuse the relievable evidences. The new method improves the reliability and rationality of fusion results compared with traditional methods.
文摘针对视频情感识别中存在运算复杂度高的缺点,提出一种基于时空局部二值模式矩(Temporal-Spatial Local Binary Pattern Moment,TSLBPM)的双模态情感识别方法。首先对视频进行预处理获得表情和姿态序列;然后对表情和姿态序列分别提取TSLBPM特征,计算测试序列与已标记的情感训练集特征间的最小欧氏距离,并将其作为独立证据来构造基本概率分配(Basic Probability Assignment,BPA);最后使用Dempster-Shafer证据理论联合规则得到情感识别结果。在双模态表情和姿态情感数据库上的实验结果表明,本文提出的时空局部二值模式矩可以快速提取视频图像的时空特征,能有效识别情感状态。与其他方法的对比实验也验证了本文融合方法的优越性。
文摘基于弹着点空间分布对目标毁伤效能的差异化影响,构建导弹命中目标不同重要区域的概率分布模型,实现对传统命中精度概念的扩展。针对导弹实打试验过程复杂、费用高、次数少的实际,采用贝叶斯方法融合多源信息,基于区域划分—分布确定—先验融合—后验求解的思路进行导弹命中精度估计。选取Dirichlet分布作为命中精度参数的先验分布,运用D-S(Dempster-Shafer)证据理论对先验信息进行融合处理,基于马尔可夫链蒙特卡罗(Markov chain Monte Carlo,MCMC)方法对精度参数的后验分布进行求解。示例表明,该方法能够细致描述导弹命中目标不同重要区域的概率,并科学融合多源命中精度先验信息,为导弹命论文拓展中精度估计方法及测试方案优化提供理论借鉴。
