【作者】 林炜；

【导师】 丁辛；

【作者基本信息】 东华大学 ， 数字化纺织， 2009， 硕士

【摘要】 织物的触感性能(传统上指手感)是纺织品生产和销售领域的关键评价指标。近年来,伴随着人们对织物及其制品舒适性追求的提高,以及以人机交互为平台的织物触感虚拟再现技术的需要,再次激发纺织界学者关注织物触觉质感的评价及其应用,尤以引入人工智能算法改进KAWABATA提出的多元、分类统计模型的预测能力为代表。然而,研究结果往往偏离预期效果,泛化能力弱。因此,有必要研究织物触觉质感的神经心理和心理物理属性,探索形成织物触感的本质。本论文将从认知心理学角度探讨织物触觉质感的空间维度及其表征,为织物触觉质感的科学定量表征提供依据。主要研究内容及成果如下:(1)织物触觉质感空间维度的确定触觉质感具有感知多维性的特征,研究的焦点在于确定感知属性空间,但迄今还没有确定组成织物多维感知属性空间的具体维度及其意义。本论文首先通过基于内隐记忆的语义描述实验,收集了大量用于表征织物触觉质感的描述符,然后在此基础上,通过聚类分析和因子分析相结合的算法,确立了表征织物触觉质感的空间维度。由聚类分析得出,滑爽—粗糙感、柔软—硬挺感以及薄—厚感可作为组成织物触觉质感属性空间的前三维,而第四维则是与粘涩感有关的维度;进一步通过因子分析证实,前五组因子的累计贡献率约达70%,即聚类分析得到的结果可解释触觉质感的主要差异。(2)触觉质感维度的辨别属性及其表征基于研究内容(1)确定的空间维度,进一步通过成对比较和大小估计相结合的心理物理实验,研究了在一定的织物表面粗糙度、弯曲刚度、物理厚度值范围内,测试者的触觉感知敏感性;并对每一维触觉质感属性感官值的分布函数和函数分布规律作了初步探索,研究发现织物触感属性的各触感心理计量函数服从正态分布。(3)由基本触觉质感属性预测织物触感基于研究内容(2)获取的组成织物触感的基本触觉质感属性,由KES测得相应的基本物理属性,采用多元线性回归分析的方法预测织物的触感,并从认知神经科学角度解释织物触感维度。本研究确立了织物触觉质感感知属性空间维度,同时分析了在一定范围内的各触感维度的分辨率,并预测了各属性维度的多元线性回归模型,从认知科学角度解释织物触感维度,有助于织物触觉质感的科学定量表征,在指导生产方面有一定的现实意义。更多还原

【Abstract】 Fabric tactile textures(namely handle) is a key evaluation for production and marketing area.In recent years,accompanied by the improvement of people’s pursuit of the fabric’s quality and comfort,as well as the needs of the virtual reproduction fabric touch technology in human-computer interaction platform, textiles’ scholars are stimulated to concerned about the tactile texture’s evaluation and application,particularly represented by the multi-classify forecast models proposed by KAWABATA,which is improved by the introduction of artificial intelligence algorithms.However,the results tend to deviate from the expected results and its generalization ability is weak.Therefore,it is necessary to study the tactile texture’s neuro-psychology and psychological-physics properties and explore the essential of the tactile texture.In this paper,from the perspective of cognitive psychology,perceptual dimensions of tactile textures and characterization are discussed which is beneficial to the quantitative evaluation of fabric hand.The main research work and results carried out is listed as following:(1) Determination of tactile perceptual dimensionsFabric tactile textures are composed of multi-dimensional perceptual attributes; the focus of the study is to identify the primary components of tactile texture space,However,it is indeterminate as well as their meanings yet.In this study,by recordings of semantic experiments on implicit memory,a lot of tactile texture descriptors are collected,and then on that basis,through cluster analysis,together with factor analysis,perceptual dimensions of tactile texture are determined.By clustering analysis,smoothness-roughness,softness-hardness as well as thinness-thickness are judged to be the first three dimensional perceptual space; and the fourth dimension is considered about the sense of stick.The results from factor analysis indicate that the cumulative contribution ratio of the five sets of factors reaches to about 70%,indicating that the results of the cluster analysis can explain the majority perceptual attributes of the tactile texture of fabrics.(2) Identification and characterization of tactile perceptual dimensionsBased on research work(1),the determination of tactile perceptual dimensions, through further physical and psychological experiments of comparison and the estimated,the sensitivities of the tactile perception are discussed on the study of the fabric’s surface roughness,bending stiffness,and physical thickness.In addition,a preliminary exploration are taken for the function and distribution of the value of textiles’ sensory properties on each touch-dimensional,which found that the properties of the fabric touch psychometric yield to normal distribution function.(3) Tactile texture forecast base on the fabric propertiesBased on the results of the composition of tactile perceptual properties from research work(2),with physical properties measured by KES test,it is using multiple linear regression analysis to forecast the perceptual textile texture,and explaining it from the perspective of cognitive neuroscience.In this thesis,the study established the tactile texture space and analysis of the identification properties in a certain scope of each property dimensions.At the same time,predicted the multiple linear regression model of the perceptual dimensions,explained the dimensions from the perspective of cognitive neuroscience,which is contributed to the science quantitative characterization of fabric the tactile texture,and significant in guiding the production of some practical.更多还原