符合认知心理的三维人脸线条化评估方法  

Evaluation Method of Line Drawings for 3D Faces Cognition

作　　者：岳双燕 王璐[1,2] 陆巧[1,2] 孟祥旭[1,2] 

机构地区：[1]山东大学计算机科学与技术学院,济南250101 [2]数字媒体技术教育部工程研究中心,济南250101

出　　处：《计算机辅助设计与图形学学报》2016年第7期1059-1067,共9页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(61472224;61472225);国家"八六三"高技术研究发展计划(2014AA01A302);山东省自主创新及成果转化专项(2014ZZCX08201);山东省重点研发计划(2015GGX106006);泰山学者建设工程专项经费;山东大学青年学者未来计划资助项目(2015WLJH41)

摘　　要：针对三维人脸几种典型的特征线条表达类型,结合人脸的领域知识和人脸识别认知机理,面向人脸线条化艺术方式,提出一种适用于人脸识别认知机理的人脸表情、朝向和五官的主观识别和评估方法.首先对大量的用户人脸进行识别测试,统计不同风格化线条在识别过程的准确率和效率,评估在不同表情和人脸区域中不同线条的表达能力;然后依据评估结论融合不同类型线条,生成人脸线条化优化绘制结果.实验结果表明,优化后的线条化结果更加有利于人脸的主观识别,该方法可推广至人脸肖像画、油画等多种风格,适用于人脸肖像具有象征主义的艺术表达形式.We present a new evaluation method to evaluate facial line drawings extracted from different 3D facial features, such as facial expressions, head orientations and face characteristics. First, our method assesses accuracies and efficiencies of different stylized lines in recognition process from face recognition user tests, and evaluates the representativeness of different lines in various facial expressions and facial regions. Then, by integrating different kinds of lines, our method generates optimized facial line drawings. Results show that these optimized facial line drawings are more conducive to subjective face recognition. Furthermore, our recognition and evaluation method can be extended to many other symbolisti art forms such as portraits and oil paintings.

关 键 词：非真实感绘制 风格化人脸 线条化 风格化绘制评估 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]