检索结果-南通市图书馆

同方期刊数据库

在线全文

学校读者我要写书评

暂无评论

基于深度神经网络的台风中心定位方法

热带气象学报 2024年第3期40卷 341-351页

作者：郑宗生沈绪坤王振华卢鹏上海海洋大学信息学院上海201306

台风中心定位的微小误差会对台风路径预报造成较大的偏离,因此精确定位台风中心是台风路径预测和灾害预报的重要步骤。台风云系随时间不断变化且风力强弱不一,在卫星云图中呈现了多样性和复杂性,现有基于神经网络的模型由于缺少对台风... 详细信息

台风中心定位的微小误差会对台风路径预报造成较大的偏离,因此精确定位台风中心是台风路径预测和灾害预报的重要步骤。台风云系随时间不断变化且风力强弱不一,在卫星云图中呈现了多样性和复杂性,现有基于神经网络的模型由于缺少对台风特征图像多维度参数的权重合理分配,在自动提取台风图像特征上受到了限制。为此,提出一种融合通道注意力与坐标注意力的神经网络模型(TY-LOCNet),首先搭建深度卷积神经网络模型提取台风特征;其次引入通道注意力机制从台风特征中捕获通道级别的信息,提升模型对重要通道的关注度;然后将通道注意力结果输入到坐标注意力机制中全局标定台风位置信息,使模型能够在较大的区域关注到台风的形态结构;此外,均方误差损失函数未能融合计算坐标导致定位精度低,因此提出距离损失函数(DISTLoss)通过距离回归提高模型定位精度。实验结果表明,TY-LOCNet的平均位置误差(MLE)、平均定位误差(MAE)和检测速度分别为3.502像素,0.292°和17 FPS,优于其他模型。台风中心定位模型TY-LOCNet可为台风预报提供实时性台风中心定位支持。

关键词：台风中心定位注意力机制神经网络距离损失函数

Reconstructing and editing fluids using the adaptive multilayer external force guiding model

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2022年第11期65卷 107-122页

作者： Xiaoying NIE Yong HU xukun shen Zhiyuan SU State Key Laboratory of Virtual Reality Technology and Systems Beihang University School of New Media Art and Design Beihang University

Reconstructing monocular fluids usually involves a tedious trial-and-error process, and editing desired fluid behaviors is notoriously difficult to predict and control. To address these problems, we propose an adaptive multilayer external force guiding model that alleviates the challenging parameter tuning and satisfies user-defined requirements. External forces cause the effect of target particles on fluid *** adaptive multilayer scheme makes the whole 3D fluid volume subject to the shape and motion of the water captured by the input video or designed by users. Therefore, we can avoid the tedious and laborious parameter tuning and easily balance the smoothness of fluid volume and the details of the water ***, to vividly reproduce the inflow and outflow of the video scene, we construct a generation and extinction model to add or delete fluid particles according to the three-dimensional velocity field of target particles calculated using a hybrid model coupling shape-from-shading with optical flow. Besides, we edit fluids by subtly extracting features selected by users using the off-screen rendering method and seamlessly integrating them using the dynamic weight approach. Experiments show that our approach is comparable to the state-of-the-art reconstruction quality and is remarkably convenient in authoring flows by editing ***, our results can be effectively applied to any desired new scenario.

关键词： reconstructing fluids editing fluids monocular videos adaptive multilayer external force guiding model generation and extinction model

Learning Semantic Concepts from Noisy Media Collection for Automatic Image Annotation

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2015年第4期24卷 790-794页

作者： TIAN Feng shen xukun School of Computer and Information Technology Northeast Petroleum University State Key Laboratory of Virtual Reality Technology and Systems Beihang University

Along with the explosive growth of images, automatic image annotation has attracted great interest of various research communities. However, despite the great progress achieved in the past two decades, automatic annotation is still an important open problem in computer vision, and can hardly achieve satisfactory performance in real-world environment. In this paper, we address the problem of annotation when noise is interfering with the dataset. A semantic neighborhood learning model on noisy media collection is proposed. Missing labels are replenished, and semantic balanced neighborhood is construct. The model allows the integration of multiple label metric learning and local nonnegative sparse *** construct semantic consistent neighborhood for each sample, thus corresponding neighbors have higher global similarity, partial correlation, conceptual similarity along with semantic balance. Meanwhile, an iterative denoising method is also proposed. The method proposed makes a marked improvement as compared to the current state-ofthe-art.

关键词： Image annotation Noisy media collection Semantic neighborhood

Image Tagging by Semantic Neighbor Learning Using User-Contributed Social Image Datasets

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Tsinghua Science and Technology 2017年第6期22卷 551-563页

作者： Feng Tian xukun shen Xianmei Liu Maojun Cao the School of Computer and Information Technology Northeast Petroleum University Daqing 163318 China xukun shen is with State Key Laboratory of Virtual Reality Technology and Systems Beihang University Beijing 100191 China.

The explosive increase in the number of images on the Internet has brought with it the great challenge of how to effectively index, retrieve, and organize these resources. Assigning proper tags to the visual content is key to the success of many applications such as image retrieval and content mining. Although recent years have witnessed many advances in image tagging, these methods have limitations when applied to high-quality and large-scale training data that are expensive to obtain. In this paper, we propose a novel semantic neighbor learning method based on user-contributed social image datasets that can be acquired from the Web＇s inexhaustible social image content. In contrast to existing image tagging approaches that rely on high-quality image-tag supervision, we acquire weak supervision of our neighbor learning method by progressive neighborhood retrieval from noisy and diverse user-contributed image collections. The retrieved neighbor images are not only visually alike and partially correlated but also semantically related. We offer a step-by-step and easy-to-use implementation for the proposed method. Extensive experimentation on several datasets demonstrates that the performance of the proposed method significantly outperforms others.

关键词： image tag social image tagging user-contributed datasets semantic neighbor learning

A method of 3D modeling and codec

在线全文

学校读者我要写书评

暂无评论

Science in China(Series F) 2009年第5期52卷 758-769页

作者： QI Yue YANG shen CAI Su HOU Fei shen xukun ZHAO QinPing State Key Laboratory of Virtual Reality Technology and Systems Beihang University Beijing 100191 China Institute of Modern Educational Technology Beijing Normal University Beijing 100875 China

3D modeling and codec of real objects are hot issues in the field of virtual reality. In this paper, we propose an automatic registration two range images method and a cycle based automatic global registration algorithm for rapidly and automatically registering all range images and constructing a realistic 3D model. Besides, to meet the requirement of huge data transmission over Internet, we present a 3D mesh encoding/decoding method for encoding geometry, topology and attribute data with high compression ratio and supporting progressive transmission. The research results have already been applied successfully in digital museum.

关键词： 3D modeling range images registration 3D model compression and codec

在线全文

学校读者我要写书评

暂无评论

弱标签环境下基于语义邻域学习的图像标注

计算机研究与发展 2014年第8期51卷 1821-1832页

作者：田枫沈旭昆虚拟现实技术与系统国家重点实验室(北京航空航天大学) 北京100191 东北石油大学计算机与信息技术学院黑龙江大庆163318

图像语义自动标注是实现图像语义检索与管理的关键,是具有挑战性的研究课题.传统的图像标注方法需要具有完整、准确标签的数据集才能取得较好的标注性能.然而,在现实应用中获得数据的标签往往是不准确、不完整的,并且标签分布不均衡.对... 详细信息

图像语义自动标注是实现图像语义检索与管理的关键,是具有挑战性的研究课题.传统的图像标注方法需要具有完整、准确标签的数据集才能取得较好的标注性能.然而,在现实应用中获得数据的标签往往是不准确、不完整的,并且标签分布不均衡.对于Web图像和社会化图像尤其如此.为了更好地利用这些弱标签样本,提出了一种基于语义邻域学习的图像自动标注方法(semantic neighborhood learning from weakly labeled image,SNLWL).首先在邻域标签损失误差最小化意义下,填充训练集样本标签.通过递进式的邻域选择过程,保证建立的语义一致邻域内样本具有全局相似性、部分相关性和语义一致性,并且语义标签分布平衡.在邻域标签重构误差最小化意义下进行标签预测,降低噪声标签对性能的影响.多个数据集上的实验结果表明,与已知的具有较好标注效果的方法相比,此方法更适用于处理弱标签数据集,标准评测集上的测试也表明了此方法的有效性.

关键词：图像标注自动标注弱标签语义邻域邻域学习

在线全文

学校读者我要写书评

暂无评论

一种手部实时跟踪与定位的虚实碰撞检测方法

计算机辅助设计与图形学学报 2011年第4期23卷 713-718页

作者：李岩沈旭昆北京航空航天大学虚拟现实技术与系统国家重点实验室北京100191

在虚实融合系统的协同工作环境中操作者需要利用手部与虚物进行交互,实时地进行虚实物体间的相交检测与碰撞响应对加强整个环境的沉浸感与真实性至关重要.文中通过多种颜色标记手部各关键点来建立映射关系,并以颜色直方图为目标检测模... 详细信息

在虚实融合系统的协同工作环境中操作者需要利用手部与虚物进行交互,实时地进行虚实物体间的相交检测与碰撞响应对加强整个环境的沉浸感与真实性至关重要.文中通过多种颜色标记手部各关键点来建立映射关系,并以颜色直方图为目标检测模式对各色块进行跟踪;然后通过双目立体视觉原理计算关键点在2幅图像间的视差,进而实时恢复出三维信息,实现手部精确定位;同时,对虚物模型划分层次二叉树并逐层建立包围盒,最终通过实时判断手部与虚物的相对位置实现了精确的碰撞检测.实验结果表明,该方法所得的碰撞检测结果在摄像头光心前2m范围内较为精准,并可达到实时流畅的应用要求.

关键词：虚实融合颜色跟踪三维定位碰撞检测

GPU-BASED FLUID SIMULATION WITH FAST COLLISION DETECTION ON BOUNDARIES

在线全文

学校读者我要写书评

暂无评论

International Journal of Modeling, Simulation, and Scientific Computing 2012年第1期3卷 137-148页

作者： JIAWEN WU FENGQUAN ZHANG xukun shen State Key Laboratory of Virtual Reality Technology and System Beihang University 100191 BeijingChina

In this paper,we present a method for fluid simulation based on smoothed particle hydrodynamic(SPH)with fast collision detection on boundaries on *** major goal of our algorithm is to get a fast SPH simulation and rendering on ***,our algorithm has the following three features:At first,to make the SPH method GPU-friendly,we introduce a spatial hash method for neighbor *** sorting the particles based on their grid index,neighbor search can be done quickly on ***,we propose a fast particle-boundary collision detection *** precomputing the distance field of scene boundaries,collision detection’s computing cost arrived as O(n),which is much faster than the traditional ***,we propose a pipeline with fine-detail surface reconstruction,and progressive photon mapping working on *** experiment our algorithm on different situations and particle numbers of scenes,and find out that our method gets good *** experimental data shows that we can simulate 100K particles,and up to 1000K particles scene at a rate of approximately 2 times per second.

关键词： SPH simulation fast collision detection GPU progressive photon mapping

基于多模态触觉与伪触觉反馈的智能设计触觉体验构建研究

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

装饰 2022年第9期 28-33页

作者：马晓娜童倩倩李一含沈旭昆北京航空航天大学新媒体艺术与设计学院鹏城实验室数学与理论部

触觉本身包括多模态的信息,本文聚焦智能设计中触觉体验的建构,分析了触觉多模态信息(力度、刚度、纹理、温度等)和多通道感官线索(触觉、视觉、听觉等)如何影响触觉感知,提出了智能设计中以触觉多模态信息构建为主,基于伪触觉反馈的多... 详细信息

触觉本身包括多模态的信息,本文聚焦智能设计中触觉体验的建构,分析了触觉多模态信息(力度、刚度、纹理、温度等)和多通道感官线索(触觉、视觉、听觉等)如何影响触觉感知,提出了智能设计中以触觉多模态信息构建为主,基于伪触觉反馈的多模态搭配与转化为辅的触觉体验的构建框架,探索触觉本身的触感反馈与力感反馈以及伪触觉反馈机制,为解决智能设计中触觉反馈的受限问题提供思路。

关键词：触觉智能交互设计多模态伪触觉