二维卡通画的渲染方法研究
发布时间：2018-10-26 19:27
【摘要】： 非真实感图形学作为计算机图形学一个新的分支,自20世纪90年代中期产生至今,已在多个领域得到了很好的发展,形成了以绘制、建模及动画为核心的理论与实践全面深入的研究和创新。其中,在模拟艺术目的的驱动下,已经产生了多种风格的非真实感渲染方法。基于图像的卡通风格渲染由于艺术风格活泼、表现形式灵活,已经吸引了许多图形学研究者致力于这项工作,这也成为本文选题的主要动机和出发点。 非真实感绘制技术可分为两类:一类是基于图像空间的二维绘制,另一类是基于物体空间的三维绘制。本文对诸多艺术渲染中的卡通风格渲染进行了深入研究,从图形学和艺术两个层面上分析了二维卡通和Q版卡通的造型特点,进而提出自己的算法,得到了相应的绘制结果。其中主要的创新性工作包括: (1)提出了基于扫描线的二维卡通风格渲染算法。对用户输入的用数码相机拍摄的图像首先提取轮廓,然后进行颜色处理,最后合并得到具有卡通风格的图像。同时,给出了基于直方图的边缘参数的计算和基于扫描线的颜色处理两个算法,自动实时地完成了渲染过程。实验结果表明该方法的渲染效果更接近于手绘风格,适合于卡通动画和个性化照片等领域的应用。 (2)提出了基于图像的Q版卡通生成算法。该方法的输入为一张真实的人物正面直立照片,输出为具有Q版卡通风格的漫画形象。主要研究内容包括:第一,人物整体和头部的夸张变形。本文采用归一化方法、并提出基于抛物样条的局部放大算法,分别实现了变形。第二,附属细节的填补与挪移。本文考虑了变形对区域外细节的影响,通过建立发型库以及基于纹理合成的旋转修正了缺陷,使得效果更加逼真。该方法设计简单,易于控制,可应用于游戏、动画、广告等领域。 (3)提出了基于抛物线的图像变形算法。首先,将变形前后的部分嵌入一个包含最小的抛物样条曲线;然后建立两条曲线内部像素之间的映射关系,获取新抛物线内部像素的坐标,以及原抛物线内部相应像素的纹理信息,进行重写,完成变形。实验证明,该变形算法对区域内的像素实施了变形,同时保证区域外像素不受影响,实现了有效的局部变形。 (4)提出了基于椭圆窗口的图像裁剪算法。首先采用高效的椭圆算法生成裁剪窗口,然后以窗口作为边界利用扫描线原理确定窗口内象素并输出,完成裁剪过程。该方法无需复杂的求交运算,只需简单的象素读写便可完成裁剪操作,因此裁剪过程简单,执行速度快,算法复杂度低,尤其适合于复杂图形及图像的裁剪处理。
[Abstract]:Non-realistic graphics, as a new branch of computer graphics, has been developed in many fields since the mid-1990s. Modeling and animation as the core of the theory and practice of comprehensive and in-depth research and innovation. Image based cartoon rendering has attracted many graphic researchers to devote themselves to this work because of its lively artistic style and flexible expression, which has become the main motive and starting point of this thesis. Non-realistic rendering techniques can be divided into two categories: one is 2D rendering based on image space and the other is 3D rendering based on object space. In this paper, the cartoon style rendering of many art rendering is deeply studied, and the modeling characteristics of 2D cartoon and Q cartoon are analyzed from the two levels of graphics and art, and their own algorithm is put forward, and the corresponding rendering results are obtained. The main innovative works are as follows: (1) A novel 2D cartoon rendering algorithm based on scanning line is proposed. Firstly, the contour is extracted from the image taken by the digital camera, then the color is processed. Finally, the cartoon style image is merged to get the cartoon image. At the same time, two algorithms of edge parameter calculation based on histogram and color processing based on scanning line are presented, and the rendering process is completed automatically and in real time. The experimental results show that the rendering effect of this method is more similar to the hand-drawn style, and it is suitable for cartoon animation and personalized photo applications. (2) an image based Q version cartoon generation algorithm is proposed. The input of this method is a real upright picture of the face of the character, and the output is a cartoon image with Q cartoon style. The main research contents include: first, the exaggerated deformation of the whole character and head. In this paper, a normalized method and a parabolic spline based local amplification algorithm are proposed to realize the deformation respectively. Second, the subsidiary details of the filling and removal. In this paper, the effect of deformation on the detail outside the region is considered, and the defect is corrected by establishing hairstyle database and rotation based on texture synthesis, which makes the effect more realistic. This method is simple and easy to control. It can be used in game, animation, advertising and other fields. (3) an image deformation algorithm based on parabola is proposed. Then the mapping relationship between the pixels of the two curves is established to obtain the coordinates of the pixels in the new parabola and the texture information of the corresponding pixels in the original parabola. Experimental results show that the algorithm can deform the pixels in the region and ensure that the pixels outside the region will not be affected, thus realizing the effective local deformation. (4) an image clipping algorithm based on elliptical window is proposed. First, the clipped window is generated by an efficient ellipse algorithm, and then the window is used as the boundary to determine the pixel in the window and output the pixel in the window by using the principle of scanning line, and the clipping process is completed. This method needs only simple pixel reading and writing to complete the clipping operation without complex intersection operation, so the process of clipping is simple, the speed of execution is high, and the complexity of the algorithm is low, which is especially suitable for cutting complex graphics and images.
【学位授予单位】：辽宁师范大学
【学位级别】：硕士
【学位授予年份】：2009
【分类号】：TP391.41
本文编号：2296722
[Abstract]:Non-realistic graphics, as a new branch of computer graphics, has been developed in many fields since the mid-1990s. Modeling and animation as the core of the theory and practice of comprehensive and in-depth research and innovation. Image based cartoon rendering has attracted many graphic researchers to devote themselves to this work because of its lively artistic style and flexible expression, which has become the main motive and starting point of this thesis. Non-realistic rendering techniques can be divided into two categories: one is 2D rendering based on image space and the other is 3D rendering based on object space. In this paper, the cartoon style rendering of many art rendering is deeply studied, and the modeling characteristics of 2D cartoon and Q cartoon are analyzed from the two levels of graphics and art, and their own algorithm is put forward, and the corresponding rendering results are obtained. The main innovative works are as follows: (1) A novel 2D cartoon rendering algorithm based on scanning line is proposed. Firstly, the contour is extracted from the image taken by the digital camera, then the color is processed. Finally, the cartoon style image is merged to get the cartoon image. At the same time, two algorithms of edge parameter calculation based on histogram and color processing based on scanning line are presented, and the rendering process is completed automatically and in real time. The experimental results show that the rendering effect of this method is more similar to the hand-drawn style, and it is suitable for cartoon animation and personalized photo applications. (2) an image based Q version cartoon generation algorithm is proposed. The input of this method is a real upright picture of the face of the character, and the output is a cartoon image with Q cartoon style. The main research contents include: first, the exaggerated deformation of the whole character and head. In this paper, a normalized method and a parabolic spline based local amplification algorithm are proposed to realize the deformation respectively. Second, the subsidiary details of the filling and removal. In this paper, the effect of deformation on the detail outside the region is considered, and the defect is corrected by establishing hairstyle database and rotation based on texture synthesis, which makes the effect more realistic. This method is simple and easy to control. It can be used in game, animation, advertising and other fields. (3) an image deformation algorithm based on parabola is proposed. Then the mapping relationship between the pixels of the two curves is established to obtain the coordinates of the pixels in the new parabola and the texture information of the corresponding pixels in the original parabola. Experimental results show that the algorithm can deform the pixels in the region and ensure that the pixels outside the region will not be affected, thus realizing the effective local deformation. (4) an image clipping algorithm based on elliptical window is proposed. First, the clipped window is generated by an efficient ellipse algorithm, and then the window is used as the boundary to determine the pixel in the window and output the pixel in the window by using the principle of scanning line, and the clipping process is completed. This method needs only simple pixel reading and writing to complete the clipping operation without complex intersection operation, so the process of clipping is simple, the speed of execution is high, and the complexity of the algorithm is low, which is especially suitable for cutting complex graphics and images.
【学位授予单位】：辽宁师范大学
【学位级别】：硕士
【学位授予年份】：2009
【分类号】：TP391.41
【引证文献】
相关硕士学位论文 前1条
1 颜彬彬;图像（视频）风格化技术研究[D];浙江工业大学;2012年
，本文编号：2296722
本文链接：https://www.wllwen.com/wenyilunwen/guanggaoshejilunwen/2296722.html
