View synthesis is an important building block in three dimension（3D） video processing and communications.Based on one or several views,view synthesis creates other views for the purpose of view prediction（for compr...View synthesis is an important building block in three dimension（3D） video processing and communications.Based on one or several views,view synthesis creates other views for the purpose of view prediction（for compression） or view rendering（for multiview-display）.The quality of view synthesis depends on how one fills the occlusion area as well as how the pixels are created.Consequently,luminance adjustment and hole filling are two key issues in view synthesis.In this paper,two views are used to produce an arbitrary virtual synthesized view.One view is merged into another view using a local luminance adjustment method,based on local neighborhood region for the calculation of adjustment coefficient.Moreover,a maximum neighborhood spreading strength hole filling method is presented to deal with the micro texture structure when the hole is being filled.For each pixel at the hole boundary,its neighborhood pixels with the maximum spreading strength direction are selected as candidates;and among them,the pixel with the maximum spreading strength is used to fill the hole from boundary to center.If there still exist disocclusion pixels after once scan,the filling process is repeated until all hole pixels are filled.Simulation results show that the proposed method is efficient,robust and achieves high performance in subjection and objection.展开更多
The existing interpretation of quantum mechanics is contrary to common sense. The existing quantum mechanical interpretation schemes are puzzling. The confusing theory is unconvincing, and needs to be amended and comp...The existing interpretation of quantum mechanics is contrary to common sense. The existing quantum mechanical interpretation schemes are puzzling. The confusing theory is unconvincing, and needs to be amended and completed. The successful interpretation program of quantum mechanics of local-realism and determinism is undoubtedly the most attractive. Preparing the interpretation program deserves to be chosen as a research goal. It is a very good premise to believe that an object particle consists of light-knot of monochromatic waves. According to this premise, the erroneous recognition about “superposition principle, wave-particle duality and uncertainty principle” can be corrected. Under this premise, above research goal is achieved by establishing, applying quantum mechanics inverse measurement theory, adhering to the principle that there must be a complete empirical chain in the derivation process of experimental conclusion, and using the side effect caused by accompanying-light to explain the diffraction experiment of object particles. Electron secondarily diffraction and other experiments directly prove that there is the measurement (observation) which may not destroy quantum coherence. The diffraction experiments of all kinds of particles show that the Keeping and playing of the coherence of moving particles in the vacuum have nothing to do with their previous experience. These are the existing experiments, to be found, that support the theory of quantum inverse measurements. The verification experiment of quantum inverse measurement is designed. The absolute superiorities of quantum inverse measurement and the new view of measurement of quantum mechanics are listed. These superiorities are that: it has the characteristics of local-realism and determinism;it is not contrary to common sense and there is no confusing place;it can predict several phenomena that cannot be predicted by other theories. A solid theoretical foundation has been laid for “correctly understanding the microscopic world” and establishment of local realism quantum mechanics.展开更多
针对基于静态场景特征进行相机位姿估计的即时定位与地图构建(SLAM:Simultaneous Localization and Mapping)技术,在其前端的特征计算和匹配的过程中易受到动态物体干扰的问题,提出了实例分割结合多视几何约束的方法,以改进视觉SLAM的...针对基于静态场景特征进行相机位姿估计的即时定位与地图构建(SLAM:Simultaneous Localization and Mapping)技术,在其前端的特征计算和匹配的过程中易受到动态物体干扰的问题,提出了实例分割结合多视几何约束的方法,以改进视觉SLAM的前端特征处理,剔除动态信息的干扰。在ORB-SLAM3(Oriented FAST and Rotated BRIEF-Simultaneous Localization and Mapping3)框架的前端,并行YOLACT++(You Only Look At CoefficienTs++)实例分割线程,将分割后的结果使用多视几何约束的方法补充检验特征点动态一致性;运用EfficientNetV2网络替换YOLACT++原来的主干网络,并使用TensorRT量化实例分割模型,以减轻算法的前端计算压力。经TUM(Technical University of Munich)数据集测试结果表明,该算法在高动态环境下的定位精度较ORB-SLAM3算法平均提升了80.6%。展开更多
针对目前已有的手工设计描述符对局部曲面几何特征描述不够全面的问题,本文提出了一种高鉴别强鲁棒的多视图几何分布特征描述符(Multi-View Geometric Distribution Signatures,MGDS)。首先,基于关键点及其邻域点构建局部参考框架(Local...针对目前已有的手工设计描述符对局部曲面几何特征描述不够全面的问题,本文提出了一种高鉴别强鲁棒的多视图几何分布特征描述符(Multi-View Geometric Distribution Signatures,MGDS)。首先,基于关键点及其邻域点构建局部参考框架(Local Reference Frame,LRF),对局部曲面进行体素化,计算三维体素的质心点分布、二维扇区轮廓特征、二维网格点密度分布以及局面曲面深度波动,生成几何特征描述符。接着,基于LRF多次旋转局部曲面,产生新的形状表示,利用质心、轮廓点、密度以及z值波动信息对旋转后的曲面进行编码。通过多个视角获取这些几何特征描述符,将它们串联成一个特征向量,得到最终的多视图几何分布特征描述符MGDS。本文在RandomView,SpaceTime,Kinect以及B3R四个数据集中不同的高斯噪声以及网格分辨率下进行实验,并与目前已有的10种描述符进行比较。与其他描述符相比,MGDS描述符的性能优于已有的局部特征描述符。实验结果表明,本文所提出的MGDS具有较好的描述性与鲁棒性,可用于三维点云的准确配准。展开更多
【目的】跨视角对象级地理定位(CVOGL)旨在卫星影像上精确定位地面街景或无人机影像所观测目标的地理位置。现有方法多聚焦于图像级匹配,通过对整张影像全局处理实现跨视角关联,缺乏对特定目标的位置编码研究,导致无法将模型的注意力引...【目的】跨视角对象级地理定位(CVOGL)旨在卫星影像上精确定位地面街景或无人机影像所观测目标的地理位置。现有方法多聚焦于图像级匹配,通过对整张影像全局处理实现跨视角关联,缺乏对特定目标的位置编码研究,导致无法将模型的注意力引导到感兴趣目标。并且由于参考图像覆盖范围的变化,查询目标在对应卫星图像中的像素占比极低,精确定位较为困难。【方法】针对以上问题,本文提出了一种基于高斯核函数与异构空间对比损失的跨视角对象级地理定位方法(Cross-View Object-Level Geo-Localization Method with Gaussian Kernel Function and Heterogeneous Spatial Contrastive Loss,GHGeo),用于精确定位感兴趣目标位置。该方法首先通过高斯核函数对查询目标进行精确位置编码,实现了对目标中心点及其分布特征的精细化建模;此外还提出了动态注意力精细化融合模块来动态加权交叉感知全局上下文与局部几何特征的空间相似性,以概率密度预测查询目标在卫星影像中的精确位置;最后通过异构空间对比损失函数来约束其训练过程,缓解跨视角特征差异。【结果】本文在CVOGL数据集进行了实验,实验结果显示:GHGeo在该数据集的“无人机-卫星”任务中,当交并比(IoU)≥25%和≥50%时定位准确率分别达到67.73%和63.00%,相较于基准方法DetGeo分别提升了5.76%和5.34%;在“街景-卫星”定位任务中,对应IoU阈值下的定位准确率分别为48.41%和45.43%的定位准确率,相较于基准方法DetGeo分别提升了2.98%和3.19%。同时与TransGeo,SAFA和VAGeo等方法在CVOGL数据集上进行对比,GHGeo则展现出了更高的定位准确性。【结论】本文方法有效提升了跨视角对象级地理定位方法的精度,为城市规划监测,应急救援调度等应用领域提供关键技术支持和精确位置信息支撑。展开更多
基金supported by the National Natural Science Foundation of China(61075013)
文摘View synthesis is an important building block in three dimension（3D） video processing and communications.Based on one or several views,view synthesis creates other views for the purpose of view prediction（for compression） or view rendering（for multiview-display）.The quality of view synthesis depends on how one fills the occlusion area as well as how the pixels are created.Consequently,luminance adjustment and hole filling are two key issues in view synthesis.In this paper,two views are used to produce an arbitrary virtual synthesized view.One view is merged into another view using a local luminance adjustment method,based on local neighborhood region for the calculation of adjustment coefficient.Moreover,a maximum neighborhood spreading strength hole filling method is presented to deal with the micro texture structure when the hole is being filled.For each pixel at the hole boundary,its neighborhood pixels with the maximum spreading strength direction are selected as candidates;and among them,the pixel with the maximum spreading strength is used to fill the hole from boundary to center.If there still exist disocclusion pixels after once scan,the filling process is repeated until all hole pixels are filled.Simulation results show that the proposed method is efficient,robust and achieves high performance in subjection and objection.
文摘The existing interpretation of quantum mechanics is contrary to common sense. The existing quantum mechanical interpretation schemes are puzzling. The confusing theory is unconvincing, and needs to be amended and completed. The successful interpretation program of quantum mechanics of local-realism and determinism is undoubtedly the most attractive. Preparing the interpretation program deserves to be chosen as a research goal. It is a very good premise to believe that an object particle consists of light-knot of monochromatic waves. According to this premise, the erroneous recognition about “superposition principle, wave-particle duality and uncertainty principle” can be corrected. Under this premise, above research goal is achieved by establishing, applying quantum mechanics inverse measurement theory, adhering to the principle that there must be a complete empirical chain in the derivation process of experimental conclusion, and using the side effect caused by accompanying-light to explain the diffraction experiment of object particles. Electron secondarily diffraction and other experiments directly prove that there is the measurement (observation) which may not destroy quantum coherence. The diffraction experiments of all kinds of particles show that the Keeping and playing of the coherence of moving particles in the vacuum have nothing to do with their previous experience. These are the existing experiments, to be found, that support the theory of quantum inverse measurements. The verification experiment of quantum inverse measurement is designed. The absolute superiorities of quantum inverse measurement and the new view of measurement of quantum mechanics are listed. These superiorities are that: it has the characteristics of local-realism and determinism;it is not contrary to common sense and there is no confusing place;it can predict several phenomena that cannot be predicted by other theories. A solid theoretical foundation has been laid for “correctly understanding the microscopic world” and establishment of local realism quantum mechanics.
文摘针对基于静态场景特征进行相机位姿估计的即时定位与地图构建(SLAM:Simultaneous Localization and Mapping)技术,在其前端的特征计算和匹配的过程中易受到动态物体干扰的问题,提出了实例分割结合多视几何约束的方法,以改进视觉SLAM的前端特征处理,剔除动态信息的干扰。在ORB-SLAM3(Oriented FAST and Rotated BRIEF-Simultaneous Localization and Mapping3)框架的前端,并行YOLACT++(You Only Look At CoefficienTs++)实例分割线程,将分割后的结果使用多视几何约束的方法补充检验特征点动态一致性;运用EfficientNetV2网络替换YOLACT++原来的主干网络,并使用TensorRT量化实例分割模型,以减轻算法的前端计算压力。经TUM(Technical University of Munich)数据集测试结果表明,该算法在高动态环境下的定位精度较ORB-SLAM3算法平均提升了80.6%。
文摘【目的】跨视角对象级地理定位(CVOGL)旨在卫星影像上精确定位地面街景或无人机影像所观测目标的地理位置。现有方法多聚焦于图像级匹配,通过对整张影像全局处理实现跨视角关联,缺乏对特定目标的位置编码研究,导致无法将模型的注意力引导到感兴趣目标。并且由于参考图像覆盖范围的变化,查询目标在对应卫星图像中的像素占比极低,精确定位较为困难。【方法】针对以上问题,本文提出了一种基于高斯核函数与异构空间对比损失的跨视角对象级地理定位方法(Cross-View Object-Level Geo-Localization Method with Gaussian Kernel Function and Heterogeneous Spatial Contrastive Loss,GHGeo),用于精确定位感兴趣目标位置。该方法首先通过高斯核函数对查询目标进行精确位置编码,实现了对目标中心点及其分布特征的精细化建模;此外还提出了动态注意力精细化融合模块来动态加权交叉感知全局上下文与局部几何特征的空间相似性,以概率密度预测查询目标在卫星影像中的精确位置;最后通过异构空间对比损失函数来约束其训练过程,缓解跨视角特征差异。【结果】本文在CVOGL数据集进行了实验,实验结果显示:GHGeo在该数据集的“无人机-卫星”任务中,当交并比(IoU)≥25%和≥50%时定位准确率分别达到67.73%和63.00%,相较于基准方法DetGeo分别提升了5.76%和5.34%;在“街景-卫星”定位任务中,对应IoU阈值下的定位准确率分别为48.41%和45.43%的定位准确率,相较于基准方法DetGeo分别提升了2.98%和3.19%。同时与TransGeo,SAFA和VAGeo等方法在CVOGL数据集上进行对比,GHGeo则展现出了更高的定位准确性。【结论】本文方法有效提升了跨视角对象级地理定位方法的精度,为城市规划监测,应急救援调度等应用领域提供关键技术支持和精确位置信息支撑。
