【作者】 刘海燕；

【导师】 徐海松；

【作者基本信息】 浙江大学 ， 光学工程， 2014， 硕士

【摘要】 颜色和纹理是物体外貌的重要组成部分,当颜色相同或相近时,纹理的结构对于人眼的感知属性有非常显著的影响。尤其是纺织品,不同的纺织纹理会使人眼的色觉感受产生巨大差异。而现有的色貌模型如CIECAM02,并不能满意地预测具有表面纹理的物体色貌。为了更好地描述纺织品的外貌属性,本文设计了一系列专门的心理物理学实验来测试评价纺织品最重要的视觉属性以及各个感知属性与计算模型之间的关系,这对纺织品在各行各业的应用具有重大意义。本文采用不同颜色不同纹理的57个纺织品样本,经过严格的颜色管理过程采集并仿真显示在专业液晶显示器(LCD)上,并基于类别判定法的心理物理学实验方法,分别对物体表面色和仿真显示色样本的15个视觉属性进行评价,最终提取出3个对视觉评价最为重要的属性,即方向性、粗糙度和温暖属性。其中,方向性和粗糙度属性与纹理结构更为相关,而温暖属性则主要依赖于物体的颜色。数据分析表明,纹理基元较大的物体方向性属性较强,但粗糙度属性也较为明显,而纹理基元模糊的纺织品恰好相反。当纺织品颜色相同时,对于表面更粗糙的样本,其对温暖属性的视觉感知略有增强。同时,物体色与仿真显示色样本的结果具有较好的一致性。基于对相关视觉属性评价结果的因子分析,采用成对比较的心理物理学实验方法,对所提取出的三个主元分量分别进行建模。实验结果表明,纺织品的方向性属性与其本身的频谱分布密切相关,即当纹理基元的排列分布情况变化时,对应的方向性属性的视觉感知也随之变化,其变化量与计算量的对数值成正比。粗糙度属性则与纺织品的空间频率和深度信息有关,当纺织品的仿真显示色样经过高斯模糊的图像平滑后,对粗糙度属性的视觉感知也随之变弱。而温暖属性的评价模型则只与纺织品显示色样的彩度和色调相关,对红、黄色样本的温暖属性视觉感知较强,对蓝、紫色样本的温暖属性感知则较弱。最后,对本文的研究工作进行了总结,并针对研究中可以扩展的方向做出了展望。更多还原

【Abstract】 Color and texture are the important parts of object appearance. The different textures would affect the visual color difference significantly when the colors of the object appearances are the same or similar. Especially in the textile industry, the texture structure has obvious effects on the visual color perceptions. But the existing color appearance model, such as CIECAM02, cannot satisfactorily predict the color appearance of the samples with textures. In order to properly describe the appearance of the textiles, a series of psychophysical experiments were specially designed to derive the most important visual properties for textiles and the computational model for each visual property, which is of great significance to the various applications of textiles.To investigate the human visual perception for the textural features of textiles,15parameters representing textile properties were evaluated in this study via the psychophysical method of categorical judgment with the surface samples presented in a light booth and simulated samples on a display, respectively. The analysis of experimental data indicates that the visual perception of textiles can be expressed by the3properties of directionality, smoothness and warmth, and directionality and smoothness properties depend on texture structure while the warmth property are highly related to the colors. The textiles with large primitive elements have high directionality and coarseness, while the textiles with unobvious primitive elements exhibit the properties of randomness and smoothness. The visual perception of the warmth property increases slightly for coarser samples when they have similar color. Meanwhile, the visual evaluations for the15properties of the surface samples coincide well with those of display samples.Based on the psychophysical method of pair comparison, three models were built respectively for the three principal visual properties. The experimental results indicate that the directionality property correlates well with the Fourier spectral distribution, i.e., when the placement rules change, its evaluation varies and is proportional to the log value of the computational directionality. The coarseness property depends on the spatial frequency and depth information, and the corresponding visual perception decreases when the display sample is made smoother through Gaussian blur. The chroma and hue of the display textile sample have a vital impact on the assessments for the warmth property, and the red and yellow samples are visually warmer than the blue and purple ones.Finally, on the basis of summarizing the research of this thesis, the prospects of the future study were also proposed.更多还原