基于复用计算的大纹理实时合成  被引量:9

Reusing Partially Synthesized Textures for Real-Time Synthesis of Large Textures

在线阅读下载全文

作  者:陈昕[1,2] 王文成[1] 

机构地区:[1]中国科学院软件研究所计算机科学国家重点实验室,北京100190 [2]中国科学院研究生院,北京100039

出  处:《计算机学报》2010年第4期768-775,共8页Chinese Journal of Computers

基  金:国家自然科学基金(60773026;60873182;60833007)资助~~

摘  要:文中提出一种基于复用计算的纹理合成方法,逐步地利用已合成的部分纹理来生成更大的纹理块,以进行后续的纹理合成计算.由此,该方法可节省大量耗时的纹理块选择及缝合计算,提高了合成效率.实验表明,新方法可实时合成2048×2048像素的大纹理,而已有工作至多只能以交互的速度进行这样的合成.Example-based texture synthesis has proved to be an effective approach to generate a perceptually similar texture from a given small input exemplar image.It has found applications in many areas such as Virtual Reality and Geography Information Systems.Till now,a large number of techniques have been proposed,being able to synthesize textures of medium sizes in real time.However,the wide requirement in applications for real time synthesis of large textures is still a big challenge to existing techniques.With regard to this,this paper proposes to reuse partially synthesized textures to speed up texture synthesis,and by a lot of tests,it can real time produce large textures in 2048×2048 pixels.By its scheme,it first produces a small part of the output texture based on general constraint computation,and then iteratively generates bigger patches from the synthesized parts of the output texture to produce new parts of the output texture.In this way,the operations on selecting patches and stitching neighboring patches can be greatly reduced for acceleration.Meanwhile,the patch sizes and the width of the overlapping zones for stitching patches are computed optimally by investigating the periodic features of the example texture,so that the new method can produce high quality textures as existing advanced techniques such as texture optimization,one of the well-known techniques for producing high quality textures.The new method is simple and robust,very useful for real time synthesis of large textures.

关 键 词:纹理合成 复用计算 实时合成 大纹理 块合成方法 

分 类 号:TP391[自动化与计算机技术—计算机应用技术]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象