基于形态美度的产品多意象预测模型  

Product Multi-Image Prediction Model Based on Aesthetic Measure of Form

作　　者：周爱民[1,2] 苏建宁[2] 阎树田[1] 欧阳晋焱[2] 张书涛[2] ZHOU Aimin;SU Jianning;YAN Shutian;OUYANG Jinyan;ZHANG Shutao(School of Mechanical&Electronical Engineering,Lanzhou University of Technology,Lanzhou Gansu 730050,China;School of Design Art,Lanzhou University of Technology,Lanzhou Gansu 730050,China)

机构地区：[1]兰州理工大学机电工程学院,甘肃兰州730050 [2]兰州理工大学设计艺术学院,甘肃兰州730050

出　　处：《图学学报》2018年第4期654-660,共7页Journal of Graphics

基　　金：国家自然科学基金项目(51465037,51705226);甘肃省自然科学基金项目(2017gs10786)

摘　　要：为了深入分析产品形态与消费者情感需求之间的关系,从认知心理学的角度,探索性地提出一种"设计特征-形态美度-感性意象"的灰箱模型,进行产品形态多意象预测。首先运用形式美学法则与计算美学理论构建产品美度指标评价体系;然后利用灰熵关联分析方法计算美度指标对多意象的影响程度,筛选主要的美度指标,避免冗余信息输入对模型预测精度的影响;最后结合各意象相互联系的特点,以主要美度指标为输入,以多意象为输出,构建多输出最小二乘支持向量回归机预测模型。利用该模型对汽车前脸3个目标意象进行了预测,结果表明其预测精度较高。For analyzing the relationship between the product form and the emotional demand of consumers,a grey box model,design features-form aesthetic measures-perceptual images,was proposed from the perspective of cognitive psychology to predict multiple images of product forms.Firstly,according to the principle of formalist aesthetics and computational aesthetics,an evaluation system of the product form aesthetic measure indexes was established.Then,the method of grey entropy association analysis was used to calculate the influence of the aesthetic measure indexes on multi-images,and screen out the main aesthetic measure indexes to reduce the influence of redundant information on prediction accuracy.Finally,according to the characteristics of the interrelation of images with each other,taking main aesthetic measure indexes as input data and multi-images as output data,the prediction model of multi-output least-squares support vector regression machines was established.Three target images about the car front faces were predicted using this model,and the results indicate its high prediction accuracy.

关 键 词：产品形态 美度 多意象 灰熵关联分析 多输出最小二乘支持向量回归机 

分 类 号：TH166[机械工程—机械制造及自动化] TB472[一般工业技术—工业设计]