CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleFace portrait generation research Based on Computer Vision
Thesis Advisor李成荣
Degree Grantor中国科学院大学
Place of Conferral中国科学院自动化研究所
Degree Discipline控制工程
Keyword计算机视觉 人脸关键点 边缘化 线条化 人脸肖像画 Computer Vision Facial Landmark Marginalization Human Face Portrait
Abstract通过计算机处理进行人脸肖像画生成是计算机视觉的研究方向之一,该技术在科普展览领域有了广泛使用,并且其中使用的计算机视觉相关的研究技术在工业生产和日常生活中也都有广泛应用。本文介绍了一种基于计算机视觉的人脸肖像画生成系统。该系统根据用户正面照片自动生成相应的肖像画线条。论文的主要研究内容和成果为: 人脸检测算法的研究。分析了图像常见颜色空间的理论,介绍了基于肤色的人脸检测方法。在综合比较了Haar-like特征的人脸检测后,采用基于Haar-like特征的人脸检测器,精确提取了人脸区域。 人脸区域关键点定位算法的研究。研究了点分布模型(Point Distribution Model, PDM)、主动形状模型(Active Shape Model, ASM)和主动外观模型(Active Appearance Model)。结合ASM模型比AAM模型计算量少,定位速度快的特点,论文采用ASM模型精确提取了人脸区域关键点,改善了人脸肖像画的整体效果。 非人脸区域外轮廓生成算法的研究。介绍了目前常用的图像分割算法,采用图像概率边缘的分割模型,对分割结果进行边缘检测,得到了非人脸区域的轮廓。 关键点线条化算法的研究。为了使生成的肖像画更加逼真,采用了贝塞尔曲线插值算法,并将图像边缘像素点进行线条化,生成若干矢量线条,供其他应用使用,如画像机器人。 该系统在CAS-PEAL人脸数据库上的正面照片进行了测试,对于不同人脸以及不同年龄阶段的人脸具有较好的鲁棒性。
Other AbstractFor face portraits generated by computer processing is one of the research direction in computer vision, with the widely used in the area of the science exhibition, and the use of computer vision technology research in industrial production and daily life. In this paper, the face portrait generation system based on computer vision generated automatically according to user’s frontal faces. The main contents and results of this paper: Researched face detection algorithm. Firstly, this paper analyzed the common color space theory. Secondly, we introduce the method of face detection based on skin color and Haar-like features face detection. Then based on Haar-like features face detection, we extracted facial region accurately. Researched face region landmark positioning algorithm. Researched Point Distribution Model (PDM), Active Shape Model (ASM), and Active Appearance Model (AAM). In combination with the ASM is faster than AAM, this paper using ASM accurately extracts the face region key points. This improves the overall effectiveness of face portrait. Researched non-face region contour generation algorithm. Introduced the algorithm of image segmentation, and by using the probability on the edge of the image segmentation model, to edge detection, obtained outline non-face region. Researched key point linearization algorithm. In order to generate more realistic portraits, we using the Bezier curve interpolation algorithm, and generating to a number of vector lines by the image edge pixel points. The system refers in this paper was tested on front images in CAS-PEAL face database. The result shows that the system is robust for different person’s face and different age range faces.
Other Identifier2011E8014661085
Document Type学位论文
Recommended Citation
GB/T 7714
申松跃. 基于计算机视觉的人脸肖像画生成研究[D]. 中国科学院自动化研究所. 中国科学院大学,2014.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_2011E801466108(9398KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[申松跃]'s Articles
Baidu academic
Similar articles in Baidu academic
[申松跃]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[申松跃]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.