基于图像距离匹配的人脸卡通化技术  

Cartoon face stylization based on image euclidean distance matching

作　　者：李倩影[1] 陈锻生[1] 吴扬扬[1] 

机构地区：[1]华侨大学计算机科学与技术学院,福建厦门361021

出　　处：《微型机与应用》2014年第10期44-46,共3页Microcomputer & Its Applications

基　　金：福建省科技计划重点项目(2011H0028);福建省科技计划重大项目(2011H6016)

摘　　要：提出了一种基于人脸匹配的卡通化方案。首先建立有代表性的人脸图像库,通过矢量绘图工具对图像库中的每幅人脸图像绘制相应的卡通图,从而建立对应的卡通图库;然后对新输入的人脸图像采用基于OpenCv训练的分类器在Viola和Jones人脸检测方法框架下检测人脸区域,通过基于图像欧氏距离的PCA人脸识别匹配得到最相似人脸,进而得到相应的卡通图。实验结果表明,该方案可以获得与输入人脸相似图像的卡通图形,而且满足现在数字娱乐的需求。This paper presents an approach to mapping a face image to a cartoon stylized graph based on image Euclidean distance matching . First , a database which contains typical faces and their cartoon figures is established , in which the corresponding cartoons are drawn as to create the cartoon gallery for the face images . Then , for an input face image , a classifier which is trained in OpenCv based on Viola＆amp;Jones ’ s face detector is used to locate the face region . And the PCA face recognition based on image Euclidean distance is applied to find a face similar face image in the database . Finally , the corresponding cartoon face is found . Experiments show that our method can get a cartoon graph for an input face image , which can be applied to meet the demands of digital entertainment .

关 键 词：卡通脸 人脸检测 人脸识别 图像欧氏距离 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]