基于简单用户交互的单张图像材质外观建模方法
发布时间：2018-08-24 14:14
【摘要】：随着计算机图形学和计算机视觉的快速发展,高级图像编辑技术近些年来发展非常迅速,越来越多的研究开始关注对图像内容的理解,如图像模型的材质和表面结构信息。本文提供了一种便捷高效的仅通过简单的用户交互从单张照片中获得材质以及外观模型的方法。对于一张在任意环境光照下的近平面材质图片,我们仅需要少量的用户交互对局部信息进行约束,就可获得模型的材质信息(包括漫反射率,镜面反射率,粗糙度),图片上各个像素的法向量信息。并且可以将计算获得的材质模型在任意给定光源下渲染。算法主要分四步:首先将输入图片上的高光和阴影部分移除,并用相似的片段补全从而将图片转变为漫反射图片;然后将获得的图片通过本征分解,分解为反射率图和阴影图,由于基础的分解算法固有的不适定性问题,难以处理具有复杂反射率变化和阴影变化的局部细节,本文在这些区域通过用户用画笔添加约束将耦合的两种信息区分出来;其次以阴影图片为输入,采用联合优化的方式同时优化法向量信息和光照信息;最后通过反射率图像以及用户提供的材质信息,对全图的各个像素进行材质建模。于此同时,我们提供了一个便捷的图形界面供用户使用,可以使用户的交互更易操作,并且可以实时查看处理的中间结果,并针对图片本身的特点对结果进行优化。实验结果显示我们的算法在处理复杂光照下的图片时,无论是在几何建模还是材质建模过程中,都可以保留较为细致的细节结果。
[Abstract]:With the rapid development of computer graphics and computer vision, advanced image editing technology has developed very rapidly in recent years. More and more research has begun to focus on the understanding of image content, such as image model material and surface structure information. This paper provides a convenient and efficient way to obtain material and appearance models from a single photo only through simple user interaction. For a picture of near-plane material under arbitrary illumination, we can obtain the material information of the model (including diffuse reflectivity, mirror reflectivity) by using only a small amount of user interaction to constrain the local information. Roughness), the normal vector information for each pixel on the picture. And the calculated material model can be rendered under any given light source. The algorithm is mainly divided into four steps: first, removing the highlights and shadows on the input images, and then converting the images into diffuse reflection images by using similar fragments, and then decomposing the obtained images into reflectivity maps and shadow images through intrinsic decomposition. Due to the inherent ill-posed problem of the basic decomposition algorithm, it is difficult to deal with the local details with complex reflectivity change and shadow variation. In this paper, two kinds of coupled information are distinguished in these areas by adding constraints with brush. Secondly, the shaded image is used as input, and the normal vector information and illumination information are optimized simultaneously by the way of joint optimization. Finally, through the reflectivity image and the material information provided by the user, the material modeling of each pixel of the whole image is carried out. At the same time, we provide a convenient graphical interface for users to use, which can make the user's interaction easier to operate, and can view the intermediate results in real time, and optimize the results according to the characteristics of the picture itself. The experimental results show that our algorithm can retain more detailed results in the process of geometric modeling and material modeling when dealing with images under complex illumination.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41
[Abstract]:With the rapid development of computer graphics and computer vision, advanced image editing technology has developed very rapidly in recent years. More and more research has begun to focus on the understanding of image content, such as image model material and surface structure information. This paper provides a convenient and efficient way to obtain material and appearance models from a single photo only through simple user interaction. For a picture of near-plane material under arbitrary illumination, we can obtain the material information of the model (including diffuse reflectivity, mirror reflectivity) by using only a small amount of user interaction to constrain the local information. Roughness), the normal vector information for each pixel on the picture. And the calculated material model can be rendered under any given light source. The algorithm is mainly divided into four steps: first, removing the highlights and shadows on the input images, and then converting the images into diffuse reflection images by using similar fragments, and then decomposing the obtained images into reflectivity maps and shadow images through intrinsic decomposition. Due to the inherent ill-posed problem of the basic decomposition algorithm, it is difficult to deal with the local details with complex reflectivity change and shadow variation. In this paper, two kinds of coupled information are distinguished in these areas by adding constraints with brush. Secondly, the shaded image is used as input, and the normal vector information and illumination information are optimized simultaneously by the way of joint optimization. Finally, through the reflectivity image and the material information provided by the user, the material modeling of each pixel of the whole image is carried out. At the same time, we provide a convenient graphical interface for users to use, which can make the user's interaction easier to operate, and can view the intermediate results in real time, and optimize the results according to the characteristics of the picture itself. The experimental results show that our algorithm can retain more detailed results in the process of geometric modeling and material modeling when dealing with images under complex illumination.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41
【相似文献】
相关期刊论文 前10条
1 孙英皓;唐棣;;一种有效的交互式图像局部内容放大方法[J];计算机工程与应用;2012年12期
2 李瑞帅;余淑华;潘凯;王圆;;图像特征金字塔快速计算方法[J];电子世界;2014年07期
3 曾喜良;王金娟;;指纹分割的块图像梯度因子聚类法[J];计算机与数字工程;2008年07期
4 刘陈;王欣欣;李凤霞;赵相坤;;一种快速保边的图像对象分割方法[J];北京理工大学学报;2010年02期
5 朱薇;刘利刚;;图像适应算法中非冗余显著图的计算[J];中国图象图形学报;2011年08期
6 柳有权;吴宗胜;韩红雷;吴恩华;;线条增强的建筑物图像抽象画生成[J];计算机辅助设计与图形学学报;2013年09期
7 吴骏,唐红梅,肖志涛,贾志成;一种基于相位信息的图像对称性检测方法[J];信号处理;2004年01期
8 邵静;高隽;赵莹;张旭东;;一种基于图像固有维度的感知物体检测方法[J];仪器仪表学报;2008年04期
9 潘如如;高卫东;;高紧度机织物图像倾斜的自动纠正[J];纺织学报;2009年10期
10 刘贵喜,赵曙光,杨万海;基于梯度塔形分解的多传感器图像融合[J];光电子·激光;2001年03期
相关会议论文 前5条
1 张一鸣;刘亚t，
本文编号：2201073
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2201073.html
