A set of algorithms is proposed in this paper to automatically transform 3D animal models to Chinese painting style. Inspired by real painting process in Chinese painting of animals, we divide the whole rendering proc...A set of algorithms is proposed in this paper to automatically transform 3D animal models to Chinese painting style. Inspired by real painting process in Chinese painting of animals, we divide the whole rendering process into two parts: borderline stroke making and interior shading. In borderline stroke making process we first find 3D model silhouettes in real-time depending on the viewing direction of a user. After retrieving silhouette information from all model edges, a stroke linking mechanism is applied to link these independent edges into a long stroke. Finally we grow a plain thin silhouette line to a stylus stroke with various widths at each control point and a 2D brush model is combined with it to simulate a Chinese painting stroke. In the interior shading pipeline, three stages are used to convert a Gouraud-shading image to a Chinese painting style image: color quantization, ink diffusion and box filtering. The color quantization stage assigns all pixels in an image into four color levels and each level represents a color layer in a Chinese painting. Ink diffusion stage is used to transfer inks and water between different levels and to grow areas in an irregular way. The box filtering stage blurs sharp borders between different levels to embellish the appearance of final interior shading image. In addition to automatic rendering, an interactive Chinese painting system which is equipped with friendly input devices can be also combined to generate more artistic Chinese painting images manually.展开更多
Existing pen and ink sketch technologies can be applied to general images, but they could not produce optimal output for images of traditional architecture, because most images consist of exquisite straight lined patt...Existing pen and ink sketch technologies can be applied to general images, but they could not produce optimal output for images of traditional architecture, because most images consist of exquisite straight lined patterns in traditional architecture, such as root tiles and window bars. The lines of roofs and eaves need to be described delicately to express pen and ink sketch most effectively. Therefore, by proposing a method to create white noise for light and shade of input images, to extract input vector directions from the white noise, and to determine the direction and length of stroke, a new expression technique is proposed for pen and ink sketch that could best reflect the characteristics of traditional architecture.展开更多
Volume visualization can not only illustrate overall distribution but also inner structure and it is an important approach for space environment research.Space environment simulation can produce several correlated var...Volume visualization can not only illustrate overall distribution but also inner structure and it is an important approach for space environment research.Space environment simulation can produce several correlated variables at the same time.However,existing compressed volume rendering methods only consider reducing the redundant information in a single volume of a specific variable,not dealing with the redundant information among these variables.For space environment volume data with multi-correlated variables,based on the HVQ-1d method we propose a further improved HVQ method by compositing variable-specific levels to reduce the redundant information among these variables.The volume data associated with each variable is divided into disjoint blocks of size 43 initially.The blocks are represented as two levels,a mean level and a detail level.The variable-specific mean levels and detail levels are combined respectively to form a larger global mean level and a larger global detail level.To both global levels,a splitting based on a principal component analysis is applied to compute initial codebooks.Then,LBG algorithm is conducted for codebook refinement and quantization.We further take advantage of progressive rendering based on GPU for real-time interactive visualization.Our method has been tested along with HVQ and HVQ-1d on high-energy proton flux volume data,including>5,>10,>30 and>50 MeV integrated proton flux.The results of our experiments prove that the method proposed in this paper pays the least cost of quality at compression,achieves a higher decompression and rendering speed compared with HVQ and provides satisficed fidelity while ensuring interactive rendering speed.展开更多
基于Microfacet理论,当光线照射到粗糙表面时会发生显著的多次镜面反射现象,而传统偏振双向反射分布函数(Polarized Bidirectional Reflectance Distribution Function,pBRDF)未能很好地描述这种现象,为了得到更精确的pBRDF模型,更好地...基于Microfacet理论,当光线照射到粗糙表面时会发生显著的多次镜面反射现象,而传统偏振双向反射分布函数(Polarized Bidirectional Reflectance Distribution Function,pBRDF)未能很好地描述这种现象,为了得到更精确的pBRDF模型,更好地分析材料的偏振特性,本文在包含镜面反射、漫反射和定向漫反射的三分量pBRDF模型基础上,进一步定义了高阶微相位角与一阶相位角的关系。改进后的pBRDF模型不仅考虑了多次镜面反射的几何衰减因子,还引入了微相位角的定义,从而建立了一个更全面的高阶镜面反射pBRDF模型。通过比较不同的pBRDF模型,并结合偏振特性采集装置进行验证,实验结果表明,本文提出的模型优于其他模型,三阶模型效果普遍优于二阶。在偏振图像渲染中,峰值信噪比和结构相似性平均分别提升10.09%和2.97%,呈现更加真实的渲染效果。验证本文提出的高阶镜面反射pBRDF模型能够更准确地描述目标表面的偏振特性。展开更多
基于深度学习的目标检测算法已广泛应用,与此同时最近的一系列研究表明现有的目标检测算法容易受到对抗性攻击的威胁,造成检测器失效.然而,聚焦于自动驾驶场景下对抗攻击的迁移性研究较少,并且鲜有研究关注该场景下对抗攻击的隐蔽性.针...基于深度学习的目标检测算法已广泛应用,与此同时最近的一系列研究表明现有的目标检测算法容易受到对抗性攻击的威胁,造成检测器失效.然而,聚焦于自动驾驶场景下对抗攻击的迁移性研究较少,并且鲜有研究关注该场景下对抗攻击的隐蔽性.针对现有研究的不足,将对抗样本的优化类比于机器学习模型的训练过程,设计了提升攻击迁移性的算法模块.并且通过风格迁移的方式和神经渲染(neural rendering)技术,提出并实现了迁移隐蔽攻击(transferable and stealthy attack,TSA)方法.具体来说,首先将对抗样本进行重复拼接,结合掩膜生成最终纹理,并将其应用于整个车辆表面.为了模拟真实的环境条件,使用物理变换函数将渲染的伪装车辆嵌入逼真的场景中.最后,通过设计的损失函数优化对抗样本.仿真实验表明,TSA方法在攻击迁移能力上超过了现有方法,并在外观上具有一定的隐蔽性.此外,通过物理域实验进一步证明了TSA方法在现实世界中能够保持有效的攻击性能.展开更多
文摘A set of algorithms is proposed in this paper to automatically transform 3D animal models to Chinese painting style. Inspired by real painting process in Chinese painting of animals, we divide the whole rendering process into two parts: borderline stroke making and interior shading. In borderline stroke making process we first find 3D model silhouettes in real-time depending on the viewing direction of a user. After retrieving silhouette information from all model edges, a stroke linking mechanism is applied to link these independent edges into a long stroke. Finally we grow a plain thin silhouette line to a stylus stroke with various widths at each control point and a 2D brush model is combined with it to simulate a Chinese painting stroke. In the interior shading pipeline, three stages are used to convert a Gouraud-shading image to a Chinese painting style image: color quantization, ink diffusion and box filtering. The color quantization stage assigns all pixels in an image into four color levels and each level represents a color layer in a Chinese painting. Ink diffusion stage is used to transfer inks and water between different levels and to grow areas in an irregular way. The box filtering stage blurs sharp borders between different levels to embellish the appearance of final interior shading image. In addition to automatic rendering, an interactive Chinese painting system which is equipped with friendly input devices can be also combined to generate more artistic Chinese painting images manually.
基金Project(2010-0021154)supported by National Research Foundation of Korea(NRF)funded by the Ministry of Education,Science and Technology through the Basic Science Research ProgramProject(2012H1B8A2025982)supported by Human Resource Training Project for Regional Innovation,Korea
文摘Existing pen and ink sketch technologies can be applied to general images, but they could not produce optimal output for images of traditional architecture, because most images consist of exquisite straight lined patterns in traditional architecture, such as root tiles and window bars. The lines of roofs and eaves need to be described delicately to express pen and ink sketch most effectively. Therefore, by proposing a method to create white noise for light and shade of input images, to extract input vector directions from the white noise, and to determine the direction and length of stroke, a new expression technique is proposed for pen and ink sketch that could best reflect the characteristics of traditional architecture.
基金the Key Research Program of the Chinese Academy of Sciences(ZDRE-KT-2021-3)。
文摘Volume visualization can not only illustrate overall distribution but also inner structure and it is an important approach for space environment research.Space environment simulation can produce several correlated variables at the same time.However,existing compressed volume rendering methods only consider reducing the redundant information in a single volume of a specific variable,not dealing with the redundant information among these variables.For space environment volume data with multi-correlated variables,based on the HVQ-1d method we propose a further improved HVQ method by compositing variable-specific levels to reduce the redundant information among these variables.The volume data associated with each variable is divided into disjoint blocks of size 43 initially.The blocks are represented as two levels,a mean level and a detail level.The variable-specific mean levels and detail levels are combined respectively to form a larger global mean level and a larger global detail level.To both global levels,a splitting based on a principal component analysis is applied to compute initial codebooks.Then,LBG algorithm is conducted for codebook refinement and quantization.We further take advantage of progressive rendering based on GPU for real-time interactive visualization.Our method has been tested along with HVQ and HVQ-1d on high-energy proton flux volume data,including>5,>10,>30 and>50 MeV integrated proton flux.The results of our experiments prove that the method proposed in this paper pays the least cost of quality at compression,achieves a higher decompression and rendering speed compared with HVQ and provides satisficed fidelity while ensuring interactive rendering speed.
文摘基于Microfacet理论,当光线照射到粗糙表面时会发生显著的多次镜面反射现象,而传统偏振双向反射分布函数(Polarized Bidirectional Reflectance Distribution Function,pBRDF)未能很好地描述这种现象,为了得到更精确的pBRDF模型,更好地分析材料的偏振特性,本文在包含镜面反射、漫反射和定向漫反射的三分量pBRDF模型基础上,进一步定义了高阶微相位角与一阶相位角的关系。改进后的pBRDF模型不仅考虑了多次镜面反射的几何衰减因子,还引入了微相位角的定义,从而建立了一个更全面的高阶镜面反射pBRDF模型。通过比较不同的pBRDF模型,并结合偏振特性采集装置进行验证,实验结果表明,本文提出的模型优于其他模型,三阶模型效果普遍优于二阶。在偏振图像渲染中,峰值信噪比和结构相似性平均分别提升10.09%和2.97%,呈现更加真实的渲染效果。验证本文提出的高阶镜面反射pBRDF模型能够更准确地描述目标表面的偏振特性。
文摘基于深度学习的目标检测算法已广泛应用,与此同时最近的一系列研究表明现有的目标检测算法容易受到对抗性攻击的威胁,造成检测器失效.然而,聚焦于自动驾驶场景下对抗攻击的迁移性研究较少,并且鲜有研究关注该场景下对抗攻击的隐蔽性.针对现有研究的不足,将对抗样本的优化类比于机器学习模型的训练过程,设计了提升攻击迁移性的算法模块.并且通过风格迁移的方式和神经渲染(neural rendering)技术,提出并实现了迁移隐蔽攻击(transferable and stealthy attack,TSA)方法.具体来说,首先将对抗样本进行重复拼接,结合掩膜生成最终纹理,并将其应用于整个车辆表面.为了模拟真实的环境条件,使用物理变换函数将渲染的伪装车辆嵌入逼真的场景中.最后,通过设计的损失函数优化对抗样本.仿真实验表明,TSA方法在攻击迁移能力上超过了现有方法,并在外观上具有一定的隐蔽性.此外,通过物理域实验进一步证明了TSA方法在现实世界中能够保持有效的攻击性能.
