CAJ | 학술논문

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。
제출료일충기우인검필배적잡통화방안。수선건립유대표성적인검도상고，통과시량회도공구대도상고중적매폭인검도상회제상응적잡통도，종이건립대응적잡통도고；연후대신수입적인검도상채용기우 OpenCv 훈련적분류기재 Viola 화 Jones 인검검측방법광가하검측인검구역，통과기우도상구씨거리적 PCA 인검식별필배득도최상사인검，진이득도상응적잡통도。실험결과표명，해방안가이획득여수입인검상사도상적잡통도형，이차만족현재수자오악적수구。
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&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 .