针对当前人体动作识别算法中由于人体躯干遮挡而导致其检测精度不佳的问题,提出了一种基于加权三视角运动历史图像耦合时序分割的动作识别算法。首先,为了有效描述动作的形状和空间分布,从视频序列中提取运动历史图像(motion history im...针对当前人体动作识别算法中由于人体躯干遮挡而导致其检测精度不佳的问题,提出了一种基于加权三视角运动历史图像耦合时序分割的动作识别算法。首先,为了有效描述动作的形状和空间分布,从视频序列中提取运动历史图像(motion history image,MHI)。随后,应用深度相机(Kinect相机)来提取深度图像,以获取人体目标的动作前景轮廓。为了识别由于身体部位造成的自我遮挡,动作前景轮廓被投影到3个视角(3V)平面,形成3V-MHI,增强了对动作的正确提取,利用3V-MHI构造了一个用于记录观测运动轨迹的MHI,克服了单视角MHI的信息局限性。然后,利用时序分割(temporal segmentation,TS),根据相邻的3V-MHI来计算动作的能量和方向的变化,以检测运动的开始和结束,从而输出运动结果。此外,计算MHI的梯度值作为每个平面对应的权重,从而得到加权3V-MHI。最后,将提取的每个直方图运动模板与预先建立的数据库进行比较,完成动作的分类识别。实验表明,该方法能有效地解决自遮挡问题,在复杂环境和光照变化下有较高的准确性与鲁棒性。展开更多
Effective mineralization of biological structures poses a significant challenge in hard tissue engineering as it necessitates overcoming geometric complexities and multistep biomineralization processes.In this regard,...Effective mineralization of biological structures poses a significant challenge in hard tissue engineering as it necessitates overcoming geometric complexities and multistep biomineralization processes.In this regard,we propose“mineral-in-shell nanoarchitectonics”,inspired by the nanostructure of matrix vesicles,which can influence multiple mineralization pathways.Our nanostructural design empowers mineral precursors with tailorable properties through encapsulating amorphous calcium phosphate within a multifunctional tannic acid(TA)and silk fibroin(SF)nanoshell.The bioinspired nanosystem facilitates efficient recruitment of mineral precursors throughout the dentin structures,followed by large-scale intradentinal mineralization both in vitro and in vivo,which provides persistent protection against external stimuli.Theoretical simulations combined with experimental studies attribute the success of intradentinal mineralization to the TA-SF nanoshell,which exhibits a strong affinity for the dentin structure,stabilizing amorphous precursors and thereby facilitating concomitant mineral formation.Overall,this bioinspired mineral-in-shell nanoarchitectonics shows a promising prospect for hard tissue repair and serves as a blueprint for next-generation biomineralization-associated materials.展开更多
传统的步态识别方法在处理自遮挡步态识别问题时,通常由于从视频序列中分割出来的轮廓有噪声而不能很好地进行特征提取。为了解决这个问题,提出了基于帧差能量图(Frame Difference Energy Image,FDEI)的遗传算法(Genetic Algorithm,GA)...传统的步态识别方法在处理自遮挡步态识别问题时,通常由于从视频序列中分割出来的轮廓有噪声而不能很好地进行特征提取。为了解决这个问题,提出了基于帧差能量图(Frame Difference Energy Image,FDEI)的遗传算法(Genetic Algorithm,GA),首先采用数学形态学图像处理方法填平轮廓的漏洞并消除噪声,然后借助于步态能量图计算出步态图像的帧差能量图,接着从轮廓图像序列中提取出步态特征,最后,利用遗传算法完成步态的识别。在中科院自动化所-B(CASIA-B)步态数据库上实验验证了所提方法的有效性,实验结果表明,与几种先进的步态方法相比,所提方法在处理自遮挡步态识别问题上取得了更好的识别效果。展开更多
提出一种基于人体皮肤多线性(Skinned Multiple Person Linear,SMPL)模型的3D动画重建方法。将SMPL模型与一副图像进行拟合,对SMPL轮廓进行扭曲,以匹配原始图像中目标人物轮廓;向投影SMPL法线贴图和蒙皮贴图应用该扭曲,获得可用于构建...提出一种基于人体皮肤多线性(Skinned Multiple Person Linear,SMPL)模型的3D动画重建方法。将SMPL模型与一副图像进行拟合,对SMPL轮廓进行扭曲,以匹配原始图像中目标人物轮廓;向投影SMPL法线贴图和蒙皮贴图应用该扭曲,获得可用于构建人物的正面视图和背面视图;将视图与拟合的3D骨架相结合,以重建目标人物轮廓。此外,针对自遮挡问题,提出一个解决方案。对包括艺术作品、海报、涂鸦等正面图进行测试,通过提出的用户接口改变3D人物姿态,通过编辑和调整,从单幅人物图像重建的3D动画接近现实动画。所提方法在细节处理方面较好,手部、头发等纹理细节部分得到了较好的保留和再现,优于一些同类方法。展开更多
A method for computing the visible regions of free-form surfaces is proposed in this paper. Our work is focused on accurately calculating the visible regions of the sequenced rational Bézier surfaces forming a so...A method for computing the visible regions of free-form surfaces is proposed in this paper. Our work is focused on accurately calculating the visible regions of the sequenced rational Bézier surfaces forming a solid model and having coincident edges but no inner-intersection among them. The proposed method calculates the silhouettes of the surfaces without tessellating them into triangle meshes commonly used in previous methods so that arbitrary precision can be obtained. The computed sil- houettes of visible surfaces are projected onto a plane orthogonal to the parallel light. Then their spatial relationship is applied to calculate the boundaries of mutual-occlusion regions. As the connectivity of the surfaces on the solid model is taken into account, a surface clustering technique is also employed and the mutual-occlusion calculation is accelerated. Experimental results showed that our method is efficient and robust, and can also handle complex shapes with arbitrary precision.展开更多
文摘针对当前人体动作识别算法中由于人体躯干遮挡而导致其检测精度不佳的问题,提出了一种基于加权三视角运动历史图像耦合时序分割的动作识别算法。首先,为了有效描述动作的形状和空间分布,从视频序列中提取运动历史图像(motion history image,MHI)。随后,应用深度相机(Kinect相机)来提取深度图像,以获取人体目标的动作前景轮廓。为了识别由于身体部位造成的自我遮挡,动作前景轮廓被投影到3个视角(3V)平面,形成3V-MHI,增强了对动作的正确提取,利用3V-MHI构造了一个用于记录观测运动轨迹的MHI,克服了单视角MHI的信息局限性。然后,利用时序分割(temporal segmentation,TS),根据相邻的3V-MHI来计算动作的能量和方向的变化,以检测运动的开始和结束,从而输出运动结果。此外,计算MHI的梯度值作为每个平面对应的权重,从而得到加权3V-MHI。最后,将提取的每个直方图运动模板与预先建立的数据库进行比较,完成动作的分类识别。实验表明,该方法能有效地解决自遮挡问题,在复杂环境和光照变化下有较高的准确性与鲁棒性。
基金support provided by the National Natural Science Foundation of China(Nos.52273135,51925304,52203180).
文摘Effective mineralization of biological structures poses a significant challenge in hard tissue engineering as it necessitates overcoming geometric complexities and multistep biomineralization processes.In this regard,we propose“mineral-in-shell nanoarchitectonics”,inspired by the nanostructure of matrix vesicles,which can influence multiple mineralization pathways.Our nanostructural design empowers mineral precursors with tailorable properties through encapsulating amorphous calcium phosphate within a multifunctional tannic acid(TA)and silk fibroin(SF)nanoshell.The bioinspired nanosystem facilitates efficient recruitment of mineral precursors throughout the dentin structures,followed by large-scale intradentinal mineralization both in vitro and in vivo,which provides persistent protection against external stimuli.Theoretical simulations combined with experimental studies attribute the success of intradentinal mineralization to the TA-SF nanoshell,which exhibits a strong affinity for the dentin structure,stabilizing amorphous precursors and thereby facilitating concomitant mineral formation.Overall,this bioinspired mineral-in-shell nanoarchitectonics shows a promising prospect for hard tissue repair and serves as a blueprint for next-generation biomineralization-associated materials.
文摘传统的步态识别方法在处理自遮挡步态识别问题时,通常由于从视频序列中分割出来的轮廓有噪声而不能很好地进行特征提取。为了解决这个问题,提出了基于帧差能量图(Frame Difference Energy Image,FDEI)的遗传算法(Genetic Algorithm,GA),首先采用数学形态学图像处理方法填平轮廓的漏洞并消除噪声,然后借助于步态能量图计算出步态图像的帧差能量图,接着从轮廓图像序列中提取出步态特征,最后,利用遗传算法完成步态的识别。在中科院自动化所-B(CASIA-B)步态数据库上实验验证了所提方法的有效性,实验结果表明,与几种先进的步态方法相比,所提方法在处理自遮挡步态识别问题上取得了更好的识别效果。
文摘提出一种基于人体皮肤多线性(Skinned Multiple Person Linear,SMPL)模型的3D动画重建方法。将SMPL模型与一副图像进行拟合,对SMPL轮廓进行扭曲,以匹配原始图像中目标人物轮廓;向投影SMPL法线贴图和蒙皮贴图应用该扭曲,获得可用于构建人物的正面视图和背面视图;将视图与拟合的3D骨架相结合,以重建目标人物轮廓。此外,针对自遮挡问题,提出一个解决方案。对包括艺术作品、海报、涂鸦等正面图进行测试,通过提出的用户接口改变3D人物姿态,通过编辑和调整,从单幅人物图像重建的3D动画接近现实动画。所提方法在细节处理方面较好,手部、头发等纹理细节部分得到了较好的保留和再现,优于一些同类方法。
文摘数据增广是提升深度学习模型性能的有效方法之一。针对多类别目标检测任务中检测性能不平衡问题,提出一种针对“短板类别”(检测性能远低于模型平均检测性能的类别)的离线数据增广方法。受Cannikin’s Law的启发,采用基于复制粘贴(copy-paste)机制的场景多样性增广方法。随机采集训练集中“短板类别”实例区域,通过相似性度量机制选取训练集中增广目标样本进行随机粘贴。为了降低随机粘贴导致的遮挡问题,采用基于自遮挡(cut-replace)机制的增广方法提升模型遮挡表达能力。通过截取样本自身区域,对特征表达最显著区域进行遮挡。实验表明,FCOS目标检测框架在PASCAL VOC数据上的平均检测精度(mean average precision,mAP)从79.10%提升到83.90%,其中短板类别更为显著,提升了20.8个百分点。在MS-COCO数据上平均检测精度提升了0.9个百分点。
基金Project supported by the National Basic Research Program (973) of China (No. 2002CB312106) and the National Natural Science Foundation of China (Nos. 60533070, and 60403047). The third author was supported by the project sponsored by a Foundation for the Author of National Excellent Doctoral Dissertation of China (No. 200342) and a Program for New Century Excellent Talents in Uni-versity (No. NCET-04-0088), China
文摘A method for computing the visible regions of free-form surfaces is proposed in this paper. Our work is focused on accurately calculating the visible regions of the sequenced rational Bézier surfaces forming a solid model and having coincident edges but no inner-intersection among them. The proposed method calculates the silhouettes of the surfaces without tessellating them into triangle meshes commonly used in previous methods so that arbitrary precision can be obtained. The computed sil- houettes of visible surfaces are projected onto a plane orthogonal to the parallel light. Then their spatial relationship is applied to calculate the boundaries of mutual-occlusion regions. As the connectivity of the surfaces on the solid model is taken into account, a surface clustering technique is also employed and the mutual-occlusion calculation is accelerated. Experimental results showed that our method is efficient and robust, and can also handle complex shapes with arbitrary precision.
