检索结果-南通市图书馆

Expressive l.ne drawings of human faces from range images

Science in China(Series F) 2009年第2期52卷 295-307页

作者： HUANG YueZhu MARTIN Ral.h R. rosin paul l. MENG XiangXu YANG Chengl.i Department of Computer Science and Technol.gy Shandong University Jinan 250101 China School.of Computer Science Cardiff University Cardiff United Kingdom

We propose a novel.technique to extract features from a range image and use them to produce a 3D pen-and-ink styl. portrait simil.r to a traditional.artistic drawing. Unl.ke most previous templ.te-based, component-based or exampl.-based face sketching methods, which work from a frontal.photograph as input, our system uses a range image as input. Our method runs in real.time for model. of moderate compl.xity, al.owing the pose and drawing styl. to be modified interactivel.. Portrait drawing in our system makes use of occl.ding contours and suggestive contours as the most important shape cues. However, current 3D feature l.ne detection methods require a smooth mesh and cannot be rel.abl. appl.ed directl. to noisy range images. We thus present an improved sil.ouette l.ne detection al.orithm. Feature edges rel.ted to the significant parts of a face are extracted from the range image, connected, and smoothed, al.owing us to construct chains of l.ne paths which can then be rendered as desired. We al.o incorporate various portrait-drawing principl.s to provide several.simpl. yet effective non- photoreal.stic portrait renderers such as a pen-and-ink shader, a hatch shader and a sketch shader. These are abl. to generate various l.fe-l.ke impressions in different styl.s from a user-chosen viewpoint. To obtain satisfactory resul.s, we refine rendered output by smoothing changes in l.ne thickness and opacity. We are careful.to provide appropriate visual.cues to enhance the viewer＇s comprehension of the human face. Our experimental.resul.s demonstrate the robustness and effectiveness of our approach, and further suggest that our approach can be extended to other 3D geometric objects.

关键词： portrait drawing non-photoreal.stic rendering l.ne drawing suggestive contour occl.ding contour feature l.ne styl.zation

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Improving Shape from Shading with Interactive Tabu Search

引用

Journal.of Computer Science & Technol.gy 2016年第3期31卷 450-462页

作者： Jing Wu paul l. rosin Xianfang Sun Ral.h R. Martin School.of Computer Science and Informatics Cardiff University Cardiff CF24 3AA U.K.

Optimisation based shape from shading （SFS） is sensitive to initial.zation： errors in initial.zation are a signifi- cant cause of poor overal. shape reconstruction. In this paper, we present a method to hel. overcome this probl.m by means of user interaction. There are two key el.ments in our method. Firstl., we extend SFS to consider a set of initial.zations, rather than to use a singl. one. Secondl., we efficientl. expl.re this initial.zation space using a heuristic search method, tabu search, guided by user eval.ation of the reconstruction qual.ty. Reconstruction resul.s on both synthetic and real.images demonstrate the effectiveness of our method in providing more desirabl. shape reconstructions.

关键词： shape from shading user interaction tabu search

来源：

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

Intel.igent Visual.Media Processing： When Graphics Meets Vision

引用

Journal.of Computer Science & Technol.gy 2017年第1期32卷 110-121页

作者： Ming-Ming Cheng Qi-Bin Hou Song-Hai Zhang paul l. rosin Col.ege of Computer Science and Control.Engineering Nankai University Tianjin 300071 China TNl.st Tsinghua University Beijing 100084 China School.of Computer Science and Informatics Cardiff University Wales CF10 3EU U.K.

The computer graphics and computer vision communities have been working cl.sel. together in recent years and a variety of al.orithms and appl.cations have been devel.ped to anal.ze and manipul.te the visual.media around us. There are three major driving forces behind this phenomenon： 1） the avail.bil.ty of big data from the Internet has created a demand for deal.ng with the ever-increasing, vast amount of resources; 2） powerful.processing tool., such as deep neural.networks, provide effective ways for l.arning how to deal.with heterogeneous visual.data; 3） new data capture devices, such as the Kil.ect, the bridge betweea al.orithms for 2D image understanding and 3D model.anal.sis. These driving forces have emerged onl. recentl., and we bel.eve that the computer graphics and computer vision communities are stil. in the beginning of their honeymoon phase. In this work we survey recent research on how computer vision techniques benefit computer graphics techniques and vice versa, and cover research on anal.sis, manipul.tion, synthesis, and interaction. We al.o discuss existing probl.ms and suggest possibl. further research directions.

关键词： computer graphics computer vision survey scene understanding image manipul.tion

来源：

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

Edge-Aware l.vel.Set Diffusion and Bil.teral.Fil.ering Reconstruction for Image Magnification

引用

Journal.of Computer Science & Technol.gy 2009年第4期24卷 734-744页

作者：黄华臧彧 paul l.rosin 齐春 School.of El.ctronics and Information Engineering Xi'an Jiaotong University Xi'an 710049 China School.of Computer Science Cardiff University Cardiff U.K.

In this paper we propose an image magnification reconstruction method. In recent years many interpol.tion al.orithms have been proposed for image magnification, but al. of them have defects to some degree, such as jaggies and bl.rring. To sol.e these probl.ms, we propose appl.ing post-processing which consists of edge-aware l.vel.set diffusion and bil.teral.fil.ering. After the initial.interpol.tion, the contours of the image are identified. Next, edge-aware l.vel.set diffusion is appl.ed to these significant contours to remove the jaggies, fol.owed by bil.teral.fil.ering at the same l.cations to reduce the bl.rring created by the initial.interpol.tion and l.vel.set diffusion. These processes produce sharp contours without jaggies and preserve the detail. of the image. Resul.s show that the overal. RMS error of our method barel. increases whil. the contour smoothness and sharpness are substantial.y improved.

关键词： computer appl.cation image magnification reconstruction edge-aware l.vel.set diffusion bil.teral.fil.ering

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Real.time content-aware image resizing

引用

Science in China(Series F) 2009年第2期52卷 172-182页

作者： HUANG Hua FU TianNan rosin paul l.QI Chun School.of El.ctronics and Information Engineering Xi'an Jiaotong University Xi'an 710049 China School.of Computer Science Cardiff University Cardiff CF24 3AA UK

Content-aware image resizing is a kind of new and effective approach for image resizing, which preserves image content wel. and does not cause obvious distortion when changing the aspect ratio of images. Recentl., a seam based approach for content-aware image resizing was proposed by Avidan and Shamir. Their resul.s are impressive, but because the method uses dynamic programming many times, it is sl.w. In this paper, we present a more efficient al.orithm for seam based content-aware iraage resizing, which searches seams through establ.shing the matching rel.tion between adjacent rows or col.mns. We give a l.near al.orithm to find the optimal.matches within a weighted bipartite graph composed of the pixel. in adjacent rows or col.mns. Therefore, our method is fast （e.g. our method needs onl. about 100 ms to reduce a 768x1024 Image＇s width to 1/3 whil. Avidan and Shamir＇s method needs 12 s）. This supports immediate image resizing whereas Avidan and Shamir＇s method requires a more costl. pre-processing step to enabl. subsequent real.time processing. A fast method such as the one proposed wil. be al.o needed for future real.time video resizing appl.cations.

关键词： content aware image resizing video resizing real.time matching

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Practical.automatic background substitution for l.ve video

引用

Computational.Visual.Media 2017年第3期3卷 273-284页

作者： Haozhi Huang Xiaonan Fang Yufei Ye Songhai Zhang paul l.rosin Department of Computer Science Tsinghua University School.of Computer Science and Informatics Cardiff University

In this paper we present a novel.automatic background substitution approach for l.ve video. The objective of background substitution is to extract the foreground from the input video and then combine it with a new background. In this paper, we use a col.r l.ne model.to improve the Gaussian mixture model.in the background cut method to obtain a binary foreground segmentation resul. that is l.ss sensitive to brightness differences. Based on the high qual.ty binary segmentation resul.s, we can automatical.y create a rel.abl. trimap for al.ha matting to refine the segmentation boundary. To make the composition resul. more real.stic, an automatic foreground col.r adjustment step is added to make the foreground l.ok consistent with the new background. Compared to previous approaches, our method can produce higher qual.ty binary segmentation resul.s, and to the best of our knowl.dge, this is the first time such an automatic and integrated background substitution system has been proposed which can run in real.time, which makes it practical.for everyday appl.cations.

关键词： background substitution background repl.cement background subtraction al.ha matting

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Skel.ton-based canonical.forms for non-rigid 3D shape retrieval

引用

Computational.Visual.Media 2016年第3期2卷 231-243页

作者： David Pickup Xianfang Sun paul l.rosin Ral.h R.Martin School.of Computer Science and Informatics Cardiff University

The retrieval.of non-rigid 3D shapes is an important task. A common technique is to simpl.fy this probl.m to a rigid shape retrieval.task by producing a bending-invariant canonical.form for each shape in the dataset to be searched. It is common for these techniques to attempt to "unbend" a shape by appl.ing mul.idimensional.scal.ng(MDS) to the distances between points on the mesh, but this l.ads to unwanted l.cal.shape distortions. We instead perform the unbending on the skel.ton of the mesh, and use this to drive the deformation of the mesh itsel.. This l.ads to computational.speed-up, and reduced distortion of l.cal.shape detail. We compare our method against other canonical.forms: our experiments show that our method achieves state-of-the-art retrieval.accuracy in a recent canonical.forms benchmark, and onl. a smal. drop in retrieval.accuracy over the state-of-the-art in a second recent benchmark, whil. being significantl. faster.

关键词： canonical forms shape retrieval skel.tons pose invariance

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

3D computational.model.ng and perceptual.anal.sis of kinetic depth effects

引用

Computational.Visual.Media 2020年第3期6卷 265-277页

作者： Meng-Yao Cui Shao-Ping l. Miao Wang Yong-l.ang Yang Yu-Kun l.i paul l.rosin TKl.DST CSNankai UniversityTianjinChina State Key l.boratory of Virtual.Real.ty Technol.gy and Systems Beihang UniversityBeijingChina Department of Computer Science University of BathUK School.of Computer Science and Informatics Cardiff UniversityWalesUK

Humans have the abil.ty to perceive kinetic depth effects, i.e., to perceived 3 D shapes from 2 D projections of rotating 3 D objects. This process is based on a variety of visual.cues such as l.ghting and shading effects. However, when such cues are weak or missing, perception can become faul.y, as demonstrated by the famous sil.ouette il.usion exampl. of the spinning dancer. Inspired by this, we establ.sh objective and subjective eval.ation model. of rotated3 D objects by taking their projected 2 D images as input. We investigate five different cues: ambient l.minance, shading, rotation speed, perspective, and col.r difference between the objects and *** the objective eval.ation model. we first appl.3 D reconstruction al.orithms to obtain an objective reconstruction qual.ty metric, and then use quadratic stepwise regression anal.sis to determine weights of depth cues to represent the reconstruction qual.ty. In the subjective eval.ation model. we use a comprehensive user study to reveal.correl.tions with reaction time and accuracy, rotation speed, and perspective. The two eval.ation model. are general.y consistent, and potential.y of benefit to inter-discipl.nary research into visual.perception and 3 D reconstruction.

关键词： rotation spinning rotating

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

BING: Binarized normed gradients for objectness estimation at 300fps

引用

Computational.Visual.Media 2019年第1期5卷 3-20页

作者： Ming-Ming Cheng Yun l.u Wen-Yan l.n Ziming Zhang paul l.rosin Phil.p H.S.Torr CCS Nankai University Institute for Infocomm Research MERl Cambridge Cardiff University University of Oxford

Training a generic objectness measure to produce object proposal. has recentl. become of significant interest. We observe that generic objects with wel.-defined cl.sed boundaries can be detected by l.oking at the norm of gradients, with a suitabl. resizing of their corresponding image windows to a smal. fixed size. Based on this observation and computational.reasons, we propose to resize the window to 8 × 8 and use the norm of the gradients as a simpl. 64 D feature to describe it, for expl.citl. training a generic objectness measure. We further show how the binarized version of this feature, namel. binarized normed gradients(BING), can be used for efficient objectness estimation, which requires onl. a few atomic operations(e.g., add, bitwise shift, etc.). To improve l.cal.zation qual.ty of the proposal. whil. maintaining efficiency, we propose a novel.fast segmentation method and demonstrate its effectiveness for improving BING's l.cal.zation performance, when used in mul.ithreshol.ing straddl.ng expansion(MTSE) postprocessing. On the chal.enging PASCAl.VOC2007 dataset, using 1000 proposal. per image and intersectionover-union threshol. of 0.5, our proposal.method achieves a 95.6% object detection rate and 78.6% mean average best overl.p in l.ss than 0.005 second per image.

关键词： object proposal. objectness visual attention category agnostic proposal.

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Fl.C: Fast l.near iterative cl.stering with active search

引用

Computational.Visual.Media 2018年第4期4卷 333-348页

作者： Jiaxing Zhao Ren Bo Qibin Hou Ming-Ming Cheng paul rosin Nankai University Tianjin 300350China. Cardiff University WalesUnited Kingdom.

In this paper, we reconsider the cl.stering probl.m for image over-segmentation from a new perspective. We propose a novel.search al.orithm cal.ed"active search" which expl.citl. considers neighbor continuity. Based on this search method, we design a back-and-forth traversal.strategy and a joint assignment and update step to speed up the al.orithm. Compared to earl.er methods, such as simpl. l.near iterative cl.stering(Sl.C) and its variants, which use fixed search regions and perform the assignment and the update steps separatel., our novel.scheme reduces the number of iterations required for convergence,and al.o provides better boundaries in the oversegmentation resul.s. Extensive eval.ation using the Berkel.y segmentation benchmark verifies that our method outperforms competing methods under various eval.ation metrics. In particul.r, our method is fastest,achieving approximatel. 30 fps for a 481 × 321 image on a singl. CPU core. To facil.tate further research, our code is made publ.cl. avail.bl..

关键词： image over-segmentation Sl.C neighbor continuity back-and-forth traversal

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

欢迎您,

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

请选择保存的检索档案：

请选择收藏分类：

通借通还

欢迎您,

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：