This paper presents a new method for extract three-dimensional （3D） discrete spherical Fourier descriptors based on surface curvature voxels for pollen particle recognition. In order to reduce the high amount of pol...This paper presents a new method for extract three-dimensional （3D） discrete spherical Fourier descriptors based on surface curvature voxels for pollen particle recognition. In order to reduce the high amount of pollen information and noise disturbance, the geometric normalized curvature voxels with the principal curvedness are first extracted to represent the intrinsic pollen volumetric data. Then the curvature voxels are decomposed into radial and angular components with spherical harmonic transform in spherical coordinates. Finally the 3D discrete Fourier transform is applied to the decomposed curvature voxels to obtain the 3D spherical Fourier descriptors for pollen recognition. Experimental results show that the presented descriptors are invariant to different pollen particle geometric transformations, such as pose change and spatial rotation, and can obtain high recognition accuracy and speed simultaneously.展开更多
The generation of high-quality 3D models from single 2D images remains challenging in terms of accuracy and completeness.Deep learning has emerged as a promising solution,offering new avenues for improvements.However,...The generation of high-quality 3D models from single 2D images remains challenging in terms of accuracy and completeness.Deep learning has emerged as a promising solution,offering new avenues for improvements.However,building models from scratch is computationally expensive and requires large datasets.This paper presents a transfer-learning-based approach for category-specific 3D reconstruction from a single 2D image.The core idea is to fine-tune a pre-trained model on specific object categories using new,unseen data,resulting in specialized versions of the model that are better adapted to reconstruct particular objects.The proposed approach utilizes a three-phase pipeline comprising image acquisition,3D reconstruction,and refinement.After ensuring the quality of the input image,a ResNet50 model is used for object recognition,directing the image to the corresponding category-specific model to generate a voxel-based representation.The voxel-based 3D model is then refined by transforming it into a detailed triangular mesh representation using the Marching Cubes algorithm and Laplacian smoothing.An experimental study,using the Pix2Vox model and the Pascal3D dataset,has been conducted to evaluate and validate the effectiveness of the proposed approach.Results demonstrate that category-specific fine-tuning of Pix2Vox significantly outperforms both the original model and the general model fine-tuned for all object categories,with substantial gains in Intersection over Union(IoU)scores.Visual assessments confirm improvements in geometric detail and surface realism.These findings indicate that combining transfer learning with category-specific fine tuning and refinement strategy of our approach leads to better-quality 3D model generation.展开更多
目的:采用基于表面的形态学分析(Surface Based Morphometry,SBM)结合基于体素的形态学分析(Voxel Based Morphology,VBM)对普通磁共振成像下脑结构未见明显异常的癫痫患者的3D-T1WI相关序列数据进行对比分析,尝试发现该类患者的脑内致...目的:采用基于表面的形态学分析(Surface Based Morphometry,SBM)结合基于体素的形态学分析(Voxel Based Morphology,VBM)对普通磁共振成像下脑结构未见明显异常的癫痫患者的3D-T1WI相关序列数据进行对比分析,尝试发现该类患者的脑内致痫灶位置,为临床诊断、治疗及预后等方面的评估提供参考依据。方法:纳入2023年7月-2024年5月在大理州第二人民医院就诊的门诊或住院无灶性成年癫痫患者34例,纳入同期健康对照组39例,采集脑部磁共振3D-T1WI序列,采用独立样本t检验对比VBM及SMB中的成年无灶性癫痫患者脑结构相关数据的差异。结果:在VBM分析中成年无灶性癫痫患者组与健康组在脑脊液、脑灰质及脑白质体积方面无差异。在SBM分析中成年无灶性癫痫患者组左侧颞上回、左侧颞下回、右侧海马的皮层厚度减小,右侧豆状壳核、左侧中央前回厚度增加。结论:相较于VBM技术,SBM技术可以作为分析癫痫患者脑结构细微变化更敏感的方法。成年无灶性癫痫患者中存在主要包括颞叶、额叶皮层厚度的形态学改变,其中具体的病理生理机制仍有待更进一步研究。展开更多
Microscale metallic structures enhanced by additive manufacturing technology have attracted extensive attention especially in microelectronics and electromechanical devices.Meniscus-confined electrodeposition(MCED)adv...Microscale metallic structures enhanced by additive manufacturing technology have attracted extensive attention especially in microelectronics and electromechanical devices.Meniscus-confined electrodeposition(MCED)advances microscale 3D metal printing,enabling simpler fabrication of superior metallic microstructures in air without complex equipment or post-processing.However,accurately predicting growth rates with current MCED techniques remain challenging,which is essential for precise structure fabrication and preventing nozzle clogging.In this work,we present a novel approach to electrochemical 3D printing that utilizes a self-adjusting,voxelated method for fabricating metallic microstructures.Diverging from conventional voxelated printing which focuses on monitoring voxel thickness for structure control,this technique adopts a holistic strategy.It ensures each voxel’s position is in alignment with the final structure by synchronizing the micropipette’s trajectory during deposition with the intended design,thus facilitating self-regulation of voxel position and reducing errors associated with environmental fluctuations in deposition parameters.The method’s ability to print micropillars with various tilt angles,high density,and helical arrays demonstrates its refined control over the deposition process.Transmission electron microscopy analysis reveals that the deposited structures,which are fabricated through layer-by-layer(voxel)printing,contain nanotwins that are widely known to enhance the material’s mechanical and electrical properties.Correspondingly,in situ scanning electron microscopy(SEM)microcompression tests confirm this enhancement,showing these structures exhibit a compressive yield strength exceeding 1 GPa.The indentation tests provided an average hardness of 3.71 GPa,which is the highest value reported in previous work using MCED.The resistivity measured by the four-point probe method was(1.95±0.01)×10^(−7)Ω·m,nearly 11 times that of bulk copper.These findings demonstrate the considerable advantage of this technique in fabricating complex metallic microstructures with enhanced mechanical properties,making it suitable for advanced applications in microsensors,microelectronics,and micro-electromechanical systems.展开更多
In this paper,a two-stage light detection and ranging(LiDAR) three-dimensional(3D) object detection framework is presented,namely point-voxel dual transformer(PV-DT3D),which is a transformer-based method.In the propos...In this paper,a two-stage light detection and ranging(LiDAR) three-dimensional(3D) object detection framework is presented,namely point-voxel dual transformer(PV-DT3D),which is a transformer-based method.In the proposed PV-DT3D,point-voxel fusion features are used for proposal refinement.Specifically,keypoints are sampled from entire point cloud scene and used to encode representative scene features via a proposal-aware voxel set abstraction module.Subsequently,following the generation of proposals by the region proposal networks(RPN),the internal encoded keypoints are fed into the dual transformer encoder-decoder architecture.In 3D object detection,the proposed PV-DT3D takes advantage of both point-wise transformer and channel-wise architecture to capture contextual information from the spatial and channel dimensions.Experiments conducted on the highly competitive KITTI 3D car detection leaderboard show that the PV-DT3D achieves superior detection accuracy among state-of-the-art point-voxel-based methods.展开更多
以腧穴解剖研究成果为基础,将临床常用的18个危险穴位的解剖结构数据融入汉堡大学VOXEL-MAN三维数字化虚拟人体中,开发一套VOXEL-MAN 3D Navigator:Acupuncture运行软件(针灸学三维影像浏览器),动态、三维显示腧穴的层次解剖结构和不同...以腧穴解剖研究成果为基础,将临床常用的18个危险穴位的解剖结构数据融入汉堡大学VOXEL-MAN三维数字化虚拟人体中,开发一套VOXEL-MAN 3D Navigator:Acupuncture运行软件(针灸学三维影像浏览器),动态、三维显示腧穴的层次解剖结构和不同角度针刺所经过的断面解剖结构,并建立相关的知识库体系,能够加深对图像内容的理解,有利于提高临床针刺疗效和避免针刺意外事故的发生,并为针灸提供一种理想直观的多媒体教学手段和方法。展开更多
An efficient voxelization algorithm is presented for polygonal models by using the hardware support for the 2 D rasterization algorithm and the GPU programmable function to satisfy the volumetric display system. The v...An efficient voxelization algorithm is presented for polygonal models by using the hardware support for the 2 D rasterization algorithm and the GPU programmable function to satisfy the volumetric display system. The volume is sampled into slices by the rendering hardware and then slices are rasterated into a series of voxels. A composed buffer is used to record encoded voxels of the target volume to reduce the graphic memory requirement. In the algorithm, dynamic vertexes and index buffers are used to improve the voxelization efficiency. Experimental results show that the algorithm is efficient for a true 3-D display system.展开更多
Now the image display techniques have made great progress. The planar display and a fully new true 3-D volumetric display technique are rapidly researched and come into the application. A method based on the voxel mak...Now the image display techniques have made great progress. The planar display and a fully new true 3-D volumetric display technique are rapidly researched and come into the application. A method based on the voxel makes the observer able to get a true 3-D effect freely without any additional facilities. This paper introduces the basic form of the swept-volume display technique and discusses its voxelization process. By the translational motion prototype, this paper emphasizes how to get the voxel mapping matrix. The translated image data are the data of the beam source deflections. Finally the voxel ordering and the optimizing are also discussed.展开更多
Recent advancements in computing research and technology will allow future immersive virtual reality systems to be voxel-based, i.e. entirely based on gap-less, spatial representations of volumetric pixels. The curren...Recent advancements in computing research and technology will allow future immersive virtual reality systems to be voxel-based, i.e. entirely based on gap-less, spatial representations of volumetric pixels. The current popularity of pixel-based videoconferencing systems could turn into true telepresence experiences that are voxel-based. Richer, non-verbal communication will be possible thanks to the three-dimensional nature of such systems. An effective telepresence experience is based on the users’ sense of copresence with others in the virtual environment and on a sense of embodiment. We investigate two main quality of service factors, namely voxel size and network latency, to identify acceptable threshold values for maintaining the copresence and embodiment experience. We present a working prototype implementation of a voxel-based telepresence system and can show that even a coarse 64 mm voxel size and an overall round-trip latency of 542 ms are sufficient to maintain copresence and embodiment experiences. We provide threshold values for noticeable, disruptive, and unbearable latencies that can serve as guidelines for future voxel and other telepresence systems.展开更多
基金Project supported by the National Natural Science Foundation of China (Grant No. 60472061)the Natural Science Foundation of Jiangsu Province,China (Grant No. BK20090149)the Natural Science Foundation of Higher Education Institutions of Jiangsu Province,China (Grant No. 08KJD520019).
文摘This paper presents a new method for extract three-dimensional （3D） discrete spherical Fourier descriptors based on surface curvature voxels for pollen particle recognition. In order to reduce the high amount of pollen information and noise disturbance, the geometric normalized curvature voxels with the principal curvedness are first extracted to represent the intrinsic pollen volumetric data. Then the curvature voxels are decomposed into radial and angular components with spherical harmonic transform in spherical coordinates. Finally the 3D discrete Fourier transform is applied to the decomposed curvature voxels to obtain the 3D spherical Fourier descriptors for pollen recognition. Experimental results show that the presented descriptors are invariant to different pollen particle geometric transformations, such as pose change and spatial rotation, and can obtain high recognition accuracy and speed simultaneously.
基金funded by the Research,Development,and Innovation Authority(RDIA)—Kingdom of Saudi Arabia—under supervision Energy,Industry,and Advanced Technologies Research Center,Taibah University,Madinah,Saudi Arabia with grant number(12979-iau-2023-TAU-R-3-1-EI-).
文摘The generation of high-quality 3D models from single 2D images remains challenging in terms of accuracy and completeness.Deep learning has emerged as a promising solution,offering new avenues for improvements.However,building models from scratch is computationally expensive and requires large datasets.This paper presents a transfer-learning-based approach for category-specific 3D reconstruction from a single 2D image.The core idea is to fine-tune a pre-trained model on specific object categories using new,unseen data,resulting in specialized versions of the model that are better adapted to reconstruct particular objects.The proposed approach utilizes a three-phase pipeline comprising image acquisition,3D reconstruction,and refinement.After ensuring the quality of the input image,a ResNet50 model is used for object recognition,directing the image to the corresponding category-specific model to generate a voxel-based representation.The voxel-based 3D model is then refined by transforming it into a detailed triangular mesh representation using the Marching Cubes algorithm and Laplacian smoothing.An experimental study,using the Pix2Vox model and the Pascal3D dataset,has been conducted to evaluate and validate the effectiveness of the proposed approach.Results demonstrate that category-specific fine-tuning of Pix2Vox significantly outperforms both the original model and the general model fine-tuned for all object categories,with substantial gains in Intersection over Union(IoU)scores.Visual assessments confirm improvements in geometric detail and surface realism.These findings indicate that combining transfer learning with category-specific fine tuning and refinement strategy of our approach leads to better-quality 3D model generation.
文摘目的:采用基于表面的形态学分析(Surface Based Morphometry,SBM)结合基于体素的形态学分析(Voxel Based Morphology,VBM)对普通磁共振成像下脑结构未见明显异常的癫痫患者的3D-T1WI相关序列数据进行对比分析,尝试发现该类患者的脑内致痫灶位置,为临床诊断、治疗及预后等方面的评估提供参考依据。方法:纳入2023年7月-2024年5月在大理州第二人民医院就诊的门诊或住院无灶性成年癫痫患者34例,纳入同期健康对照组39例,采集脑部磁共振3D-T1WI序列,采用独立样本t检验对比VBM及SMB中的成年无灶性癫痫患者脑结构相关数据的差异。结果:在VBM分析中成年无灶性癫痫患者组与健康组在脑脊液、脑灰质及脑白质体积方面无差异。在SBM分析中成年无灶性癫痫患者组左侧颞上回、左侧颞下回、右侧海马的皮层厚度减小,右侧豆状壳核、左侧中央前回厚度增加。结论:相较于VBM技术,SBM技术可以作为分析癫痫患者脑结构细微变化更敏感的方法。成年无灶性癫痫患者中存在主要包括颞叶、额叶皮层厚度的形态学改变,其中具体的病理生理机制仍有待更进一步研究。
基金supported in part by National Key R&D Program of China under Grant 2023YFB4705600in part by the National Natural Science Foundation of China under Grants 61925304,62127810 and 62203138+1 种基金in part by the National Postdoctoral Program for Innovative Talents under Grant BX20200107in part by the Self-Planned Task(No.SKLRS202205C)of State Key Laboratory of Robotics and System(HIT).
文摘Microscale metallic structures enhanced by additive manufacturing technology have attracted extensive attention especially in microelectronics and electromechanical devices.Meniscus-confined electrodeposition(MCED)advances microscale 3D metal printing,enabling simpler fabrication of superior metallic microstructures in air without complex equipment or post-processing.However,accurately predicting growth rates with current MCED techniques remain challenging,which is essential for precise structure fabrication and preventing nozzle clogging.In this work,we present a novel approach to electrochemical 3D printing that utilizes a self-adjusting,voxelated method for fabricating metallic microstructures.Diverging from conventional voxelated printing which focuses on monitoring voxel thickness for structure control,this technique adopts a holistic strategy.It ensures each voxel’s position is in alignment with the final structure by synchronizing the micropipette’s trajectory during deposition with the intended design,thus facilitating self-regulation of voxel position and reducing errors associated with environmental fluctuations in deposition parameters.The method’s ability to print micropillars with various tilt angles,high density,and helical arrays demonstrates its refined control over the deposition process.Transmission electron microscopy analysis reveals that the deposited structures,which are fabricated through layer-by-layer(voxel)printing,contain nanotwins that are widely known to enhance the material’s mechanical and electrical properties.Correspondingly,in situ scanning electron microscopy(SEM)microcompression tests confirm this enhancement,showing these structures exhibit a compressive yield strength exceeding 1 GPa.The indentation tests provided an average hardness of 3.71 GPa,which is the highest value reported in previous work using MCED.The resistivity measured by the four-point probe method was(1.95±0.01)×10^(−7)Ω·m,nearly 11 times that of bulk copper.These findings demonstrate the considerable advantage of this technique in fabricating complex metallic microstructures with enhanced mechanical properties,making it suitable for advanced applications in microsensors,microelectronics,and micro-electromechanical systems.
基金supported by the Natural Science Foundation of China (No.62103298)the South African National Research Foundation (Nos.132797 and 137951)。
文摘In this paper,a two-stage light detection and ranging(LiDAR) three-dimensional(3D) object detection framework is presented,namely point-voxel dual transformer(PV-DT3D),which is a transformer-based method.In the proposed PV-DT3D,point-voxel fusion features are used for proposal refinement.Specifically,keypoints are sampled from entire point cloud scene and used to encode representative scene features via a proposal-aware voxel set abstraction module.Subsequently,following the generation of proposals by the region proposal networks(RPN),the internal encoded keypoints are fed into the dual transformer encoder-decoder architecture.In 3D object detection,the proposed PV-DT3D takes advantage of both point-wise transformer and channel-wise architecture to capture contextual information from the spatial and channel dimensions.Experiments conducted on the highly competitive KITTI 3D car detection leaderboard show that the PV-DT3D achieves superior detection accuracy among state-of-the-art point-voxel-based methods.
文摘以腧穴解剖研究成果为基础,将临床常用的18个危险穴位的解剖结构数据融入汉堡大学VOXEL-MAN三维数字化虚拟人体中,开发一套VOXEL-MAN 3D Navigator:Acupuncture运行软件(针灸学三维影像浏览器),动态、三维显示腧穴的层次解剖结构和不同角度针刺所经过的断面解剖结构,并建立相关的知识库体系,能够加深对图像内容的理解,有利于提高临床针刺疗效和避免针刺意外事故的发生,并为针灸提供一种理想直观的多媒体教学手段和方法。
文摘An efficient voxelization algorithm is presented for polygonal models by using the hardware support for the 2 D rasterization algorithm and the GPU programmable function to satisfy the volumetric display system. The volume is sampled into slices by the rendering hardware and then slices are rasterated into a series of voxels. A composed buffer is used to record encoded voxels of the target volume to reduce the graphic memory requirement. In the algorithm, dynamic vertexes and index buffers are used to improve the voxelization efficiency. Experimental results show that the algorithm is efficient for a true 3-D display system.
文摘Now the image display techniques have made great progress. The planar display and a fully new true 3-D volumetric display technique are rapidly researched and come into the application. A method based on the voxel makes the observer able to get a true 3-D effect freely without any additional facilities. This paper introduces the basic form of the swept-volume display technique and discusses its voxelization process. By the translational motion prototype, this paper emphasizes how to get the voxel mapping matrix. The translated image data are the data of the beam source deflections. Finally the voxel ordering and the optimizing are also discussed.
文摘Recent advancements in computing research and technology will allow future immersive virtual reality systems to be voxel-based, i.e. entirely based on gap-less, spatial representations of volumetric pixels. The current popularity of pixel-based videoconferencing systems could turn into true telepresence experiences that are voxel-based. Richer, non-verbal communication will be possible thanks to the three-dimensional nature of such systems. An effective telepresence experience is based on the users’ sense of copresence with others in the virtual environment and on a sense of embodiment. We investigate two main quality of service factors, namely voxel size and network latency, to identify acceptable threshold values for maintaining the copresence and embodiment experience. We present a working prototype implementation of a voxel-based telepresence system and can show that even a coarse 64 mm voxel size and an overall round-trip latency of 542 ms are sufficient to maintain copresence and embodiment experiences. We provide threshold values for noticeable, disruptive, and unbearable latencies that can serve as guidelines for future voxel and other telepresence systems.
