Augmented reality is the merging of synthetic sensory information into a user's perception of a real environment. As one of the most important tasks in augmented scene modeling, terrain simplification research has...Augmented reality is the merging of synthetic sensory information into a user's perception of a real environment. As one of the most important tasks in augmented scene modeling, terrain simplification research has gained more and more attention. In this paper, we mainly focus on point selection problem in terrain simplification using triangulated irregular network. Based on the analysis and comparison of traditional importance measures for each input point, we put forward a new importance measure based on local entropy. The results demonstrate that the local entropy criterion has a better performance than any traditional methods. In addition, it can effectively conquer the 'short-sight' problem associated with the traditional methods.展开更多
This paper presents a method for structured scene modeling using micro stereo vision system with large field of view. The proposed algorithm includes edge detection with Canny detector, line fitting with principle axi...This paper presents a method for structured scene modeling using micro stereo vision system with large field of view. The proposed algorithm includes edge detection with Canny detector, line fitting with principle axis based approach, finding corresponding lines using feature based matching method, and 3D line depth computation.展开更多
Creating proper 3D models plays an important pole in the development of scene simulation system based on multigen creator and Vega software. However, it is very difficult to construct complex structures by multigen cr...Creating proper 3D models plays an important pole in the development of scene simulation system based on multigen creator and Vega software. However, it is very difficult to construct complex structures by multigen creator. In this paper, an approach is proposed which is utilizing 3dsmax as assistant modeling software. 3D models developed in 3dsmax could be saved in 3ds format and then imported into multigen creator software. The models are revised and then saved in fit format by creator. For reducing model＇s data, simplification strategy is proposed. The problem of constructing complex models in creator is solved smoothly. In the development of digital rocket simulation project, the models constructed by this method have good visual effect, small size, and could be driven by Vega correctly.展开更多
For target detection algorithm under global motion scene, this paper suggests a target detection algorithm based on motion attention fusion model. Firstly, the motion vector field is pre-processed by accumulation and ...For target detection algorithm under global motion scene, this paper suggests a target detection algorithm based on motion attention fusion model. Firstly, the motion vector field is pre-processed by accumulation and median filter;Then, according to the temporal and spatial character of motion vector, the attention fusion model is defined, which is used to detect moving target;Lastly, the edge of video moving target is made exactly by morphologic operation and edge tracking algorithm. The experimental results of different global motion video sequences show the proposed algorithm has a better veracity and speedup than other algorithm.展开更多
The increasing scale and complexity of 3D scene design work urge an efficient way to understand the design in multi-disciplinary team and exploit the experiences and underlying knowledge in previous works for reuse.Ho...The increasing scale and complexity of 3D scene design work urge an efficient way to understand the design in multi-disciplinary team and exploit the experiences and underlying knowledge in previous works for reuse.However the previous researches lack of concerning on relationship maintaining and design reuse in knowledge level.We propose a novel semantic driven design reuse system,including a property computation algorithm that enables our system to compute the properties while modeling process to maintain the semantic consistency,and a vertex statics based algorithm that enables the system to recognize scene design pattern as universal semantic model for the same type of scenes.With the universal semantic model,the system conducts the modeling process of future design works by suggestions and constraints on operation.The proposed framework empowers the reuse of 3D scene design on both model level and knowledge level.展开更多
目的随着电影内容的复杂化与多样化,电影场景分割成为理解影片结构和支持多媒体应用的重要任务。为提升镜头特征提取和特征关联的有效性,增强镜头序列的上下文感知能力,提出一种混合架构电影场景分割方法(hybrid architecture scene seg...目的随着电影内容的复杂化与多样化,电影场景分割成为理解影片结构和支持多媒体应用的重要任务。为提升镜头特征提取和特征关联的有效性,增强镜头序列的上下文感知能力,提出一种混合架构电影场景分割方法(hybrid architecture scene segmentation network,HASSNet)。方法首先,采用预训练结合微调策略,在大量无场景标签的电影数据上进行无监督预训练,使模型学习有效的镜头特征表示和关联特性,然后在有场景标签的数据上进行微调训练,进一步提升模型性能;其次,模型架构上混合了状态空间模型和自注意力机制模型,分别设计Shot Mamba镜头特征提取模块和Scene Transformer特征关联模块,Shot Mamba通过对镜头图像分块建模提取有效特征表示,Scene Transformer则通过注意力机制对不同镜头特征进行关联建模;最后,采用3种无监督损失函数进行预训练,提升模型在镜头特征提取和关联上的性能,并使用Focal Loss损失函数进行微调,以改善由于类别不平衡导致的精度不足问题。结果实验结果表明,HASSNet在3个数据集上显著提升了场景分割的精度,在典型电影场景分割数据集MovieNet中,与先进的场景分割方法相比,AP(average precision)、mIoU(mean intersection over union)、AUC-ROC(area under the receiver operating characteristic curve)和F1分别提升1.66%、10.54%、0.21%和16.83%,验证了本文提出的HASSNet方法可以有效提升场景边界定位的准确性。结论本文提出的HASSNet方法有效结合了预训练与微调策略,借助混合状态空间模型和自注意力机制模型的特点,增强了镜头的上下文感知能力,使电影场景分割的结果更加准确。展开更多
移动机器人在嵌入式平台上的实时精细场景理解极具挑战。文章提出了一种融合改进YOLOv5与RTAB-Map的语义同步定位与地图构建(Simultaneous Localization and Mapping,SLAM)方案。通过引入GhostNet与坐标注意力机制对YOLOv5进行轻量化改...移动机器人在嵌入式平台上的实时精细场景理解极具挑战。文章提出了一种融合改进YOLOv5与RTAB-Map的语义同步定位与地图构建(Simultaneous Localization and Mapping,SLAM)方案。通过引入GhostNet与坐标注意力机制对YOLOv5进行轻量化改进,在降低复杂度的同时增强对小尺度目标的特征提取能力。系统将改进模型的检测结果与RTAB-Map生成的稠密点云进行配准,动态构建二维语义占据网格地图。实验结果表明,改进模型平均精度达87.5%,推理速度较基准提升约40%;整套系统能以约20 fps的帧率稳定运行,生成的地图能准确标识物体语义与位置。该系统有效平衡了精度与速度,为资源受限的移动机器人实现实时环境感知提供了可行解决方案。展开更多
Dynamic infrared scene simulation is for discovering and solving the problems encountered in designing, developing and manufacturing infrared imaging guidance weapons. The infrared scene simulation is explored by usin...Dynamic infrared scene simulation is for discovering and solving the problems encountered in designing, developing and manufacturing infrared imaging guidance weapons. The infrared scene simulation is explored by using the digital grayscale modulation method. The infrared image modulation model of a digital micro-mirror device （DMD） is established and then the infrared scene simulator prototype which is based on DMD grayscale modulation is developed. To evaluate its main parameters such as resolution, contrast, minimum temperature difference, gray scale, various DMD subsystems such as signal decoding, image normalization, synchronization drive, pulse width modulation （PWM） and DMD chips are designed. The infrared scene simulator is tested on a certain infrared missile seeker. The test results show preliminarily that the infrared scene simulator has high gray scale, small geometrical distortion and highly resolvable imaging resolution and contrast and yields high-fidelity images, thus being able to meet the requirements for the infrared scene simulation inside a laboratory.展开更多
基金This paper is supported by the State Key Laboratory for Image Processing & Intelligent Control (No. TKLJ9903) National Defe
文摘Augmented reality is the merging of synthetic sensory information into a user's perception of a real environment. As one of the most important tasks in augmented scene modeling, terrain simplification research has gained more and more attention. In this paper, we mainly focus on point selection problem in terrain simplification using triangulated irregular network. Based on the analysis and comparison of traditional importance measures for each input point, we put forward a new importance measure based on local entropy. The results demonstrate that the local entropy criterion has a better performance than any traditional methods. In addition, it can effectively conquer the 'short-sight' problem associated with the traditional methods.
文摘This paper presents a method for structured scene modeling using micro stereo vision system with large field of view. The proposed algorithm includes edge detection with Canny detector, line fitting with principle axis based approach, finding corresponding lines using feature based matching method, and 3D line depth computation.
文摘Creating proper 3D models plays an important pole in the development of scene simulation system based on multigen creator and Vega software. However, it is very difficult to construct complex structures by multigen creator. In this paper, an approach is proposed which is utilizing 3dsmax as assistant modeling software. 3D models developed in 3dsmax could be saved in 3ds format and then imported into multigen creator software. The models are revised and then saved in fit format by creator. For reducing model＇s data, simplification strategy is proposed. The problem of constructing complex models in creator is solved smoothly. In the development of digital rocket simulation project, the models constructed by this method have good visual effect, small size, and could be driven by Vega correctly.
文摘For target detection algorithm under global motion scene, this paper suggests a target detection algorithm based on motion attention fusion model. Firstly, the motion vector field is pre-processed by accumulation and median filter;Then, according to the temporal and spatial character of motion vector, the attention fusion model is defined, which is used to detect moving target;Lastly, the edge of video moving target is made exactly by morphologic operation and edge tracking algorithm. The experimental results of different global motion video sequences show the proposed algorithm has a better veracity and speedup than other algorithm.
基金the National Natural Science Foundation of China(Nos.61073086 and 70871078)the National High Technology Research and Development Program (863) of China(No.2008AA04Z126)
文摘The increasing scale and complexity of 3D scene design work urge an efficient way to understand the design in multi-disciplinary team and exploit the experiences and underlying knowledge in previous works for reuse.However the previous researches lack of concerning on relationship maintaining and design reuse in knowledge level.We propose a novel semantic driven design reuse system,including a property computation algorithm that enables our system to compute the properties while modeling process to maintain the semantic consistency,and a vertex statics based algorithm that enables the system to recognize scene design pattern as universal semantic model for the same type of scenes.With the universal semantic model,the system conducts the modeling process of future design works by suggestions and constraints on operation.The proposed framework empowers the reuse of 3D scene design on both model level and knowledge level.
文摘目的随着电影内容的复杂化与多样化,电影场景分割成为理解影片结构和支持多媒体应用的重要任务。为提升镜头特征提取和特征关联的有效性,增强镜头序列的上下文感知能力,提出一种混合架构电影场景分割方法(hybrid architecture scene segmentation network,HASSNet)。方法首先,采用预训练结合微调策略,在大量无场景标签的电影数据上进行无监督预训练,使模型学习有效的镜头特征表示和关联特性,然后在有场景标签的数据上进行微调训练,进一步提升模型性能;其次,模型架构上混合了状态空间模型和自注意力机制模型,分别设计Shot Mamba镜头特征提取模块和Scene Transformer特征关联模块,Shot Mamba通过对镜头图像分块建模提取有效特征表示,Scene Transformer则通过注意力机制对不同镜头特征进行关联建模;最后,采用3种无监督损失函数进行预训练,提升模型在镜头特征提取和关联上的性能,并使用Focal Loss损失函数进行微调,以改善由于类别不平衡导致的精度不足问题。结果实验结果表明,HASSNet在3个数据集上显著提升了场景分割的精度,在典型电影场景分割数据集MovieNet中,与先进的场景分割方法相比,AP(average precision)、mIoU(mean intersection over union)、AUC-ROC(area under the receiver operating characteristic curve)和F1分别提升1.66%、10.54%、0.21%和16.83%,验证了本文提出的HASSNet方法可以有效提升场景边界定位的准确性。结论本文提出的HASSNet方法有效结合了预训练与微调策略,借助混合状态空间模型和自注意力机制模型的特点,增强了镜头的上下文感知能力,使电影场景分割的结果更加准确。
文摘移动机器人在嵌入式平台上的实时精细场景理解极具挑战。文章提出了一种融合改进YOLOv5与RTAB-Map的语义同步定位与地图构建(Simultaneous Localization and Mapping,SLAM)方案。通过引入GhostNet与坐标注意力机制对YOLOv5进行轻量化改进,在降低复杂度的同时增强对小尺度目标的特征提取能力。系统将改进模型的检测结果与RTAB-Map生成的稠密点云进行配准,动态构建二维语义占据网格地图。实验结果表明,改进模型平均精度达87.5%,推理速度较基准提升约40%;整套系统能以约20 fps的帧率稳定运行,生成的地图能准确标识物体语义与位置。该系统有效平衡了精度与速度,为资源受限的移动机器人实现实时环境感知提供了可行解决方案。
基金co-supported by China Postdoctoral Science Foundation (20090461314)
文摘Dynamic infrared scene simulation is for discovering and solving the problems encountered in designing, developing and manufacturing infrared imaging guidance weapons. The infrared scene simulation is explored by using the digital grayscale modulation method. The infrared image modulation model of a digital micro-mirror device （DMD） is established and then the infrared scene simulator prototype which is based on DMD grayscale modulation is developed. To evaluate its main parameters such as resolution, contrast, minimum temperature difference, gray scale, various DMD subsystems such as signal decoding, image normalization, synchronization drive, pulse width modulation （PWM） and DMD chips are designed. The infrared scene simulator is tested on a certain infrared missile seeker. The test results show preliminarily that the infrared scene simulator has high gray scale, small geometrical distortion and highly resolvable imaging resolution and contrast and yields high-fidelity images, thus being able to meet the requirements for the infrared scene simulation inside a laboratory.
