System development of a 3-dimensional fashion・factory・boutique (Part III)

我们尝试使用3D身体测量系统获取中老年女性个人身体图像服装的销售和购买信息，并研究女性对3D服装模拟图像的感受。结果如下：1)我们开发了一种从人体表面形状生成同构网格的方法。同构网格中三角形元素的顶点由几何规则确定。这些规则应用于已经测量过的数百个身体表面形状，但是我们无法生成任何网格。通过这次试验，对规则中的缺陷进行了修改和完善。因此，从正确测量的人体表面形状生成同构网格成为可能。在三维人体测量系统的开发中，对标记数据网格构造器进行了许多改进，以提高精度和易于操作。我们对三维查看器（DspMsh）进行了改造，使其能够同时显示和操作多个网格，自动安排构件的构造，最终呈现数据网格、同构网格和曲率表示。为了防止网格三角形的退化和逆转有时会影响展开过程的完成，还从数学模型上对三维曲面展开以形成二维壳纸图案的方法进行了修正。2)将物质表面的光谱反射率与光源在每个波长处的光谱发射率相乘，经验地再现了从物质表面计算得到的反射光谱。利用颜色匹配函数对模拟反射光谱进行色度计算。它们与所发现的值吻合得很好。3)女性模特（n=72）的3d体型图像由其他女性（n=81）评价。采用聚类分析方法提取了6种三维体型。A：略显苗条和女性化的形象，B: Y-20的理想和体重过轻的形象，C: E-70的小胸围，驼背和大肚子的形象，D：超重的形象，E：略显超重和女性化的形象，F: M- 40的略宽，方肩和男性化的形象。4)选择图像关键词进行服装三维仿真。首先，从日本136本女性时尚书籍（2000 ~ 2004年）的1075个词汇中选出136个形象词汇。秒)。这136个词在问卷调查中使用5个主要的图像关键词组进行分类（n=687，年龄在20 - 70岁之间的女性受访者）。风格1；女士与女性形象，风格2；锐利现代形象，风格3；休闲形象风格，风格4；标准和复古（老式）图像，风格5；5)在3个3d人体模型（B、C、F型20 - 70岁3岁）上模拟相同的20幅设计服装图像以及设计图像标注的决策因素，并在电脑屏幕上显示。对每个3D模拟进行评估（n=120，年龄在20 - 80岁的女性）。设计形象因素以5个主要形象关键词、38个设计形象项目和35个生活方式项目为标准进行分类。款式1：裙装、粉色质感等，款式2：类皮肤质感黑色裤装等，款式3：天蓝色类棉质t恤、深蓝色牛仔裤，款式4：柔软质感连体裙，款式5：白色蕾丝类棉质裙装、西装裙等。评价者的年龄差异对服装形象的评价影响不大。少

We tried to obtain information on selling and buying garments for individual body images of middle & elderly ages women's using a 3D body measurement system and research how the women feel about the images of 3D garment simulations. The results were as follows. 1) We have developed a process to generate an isomorphic mesh from a human body surface shape. The vertices of the triangular elements of an isomorphic mesh were determined by geometrical rules. The rules were applied to the hundreds of body surface shapes that had been measured, however we could not generate any meshes. From this trial, defects in the rules were revised and improved. In the consequence, it became possible to generate isomorphic meshes from a human body surface shape that was measured correctly. For 3D body measurement system development, many improvements were implemented in the marker data mesh constructor for more accuracy and easy of operation. We remade the 3D viewer (DspMsh) to support the display and oper … More ation of many meshes at one time, automatically arranging the component construction of the final presentation of the data mesh, isomorphic mesh and curvature representation. The method of expanding the 3D surface to develop the 2D shell paper pattern was also revised from its mathematical model to prevent the degeneration and reversion of the mesh triangles, which sometimes affected the completion of the expanding process. 2) The reflection spectrum calculated from the surface of the matter has been reproduced empirically by multiplying the surface spectral reflectance of the matter and the spectral emissivity of the light source at each wave length. Furthermore the chromaticity was calculated from the simulated reflection spectrum by applying the color matching functions. They corresponded well with the values found. 3) The 3D-body shape images of female models (n=72) were evaluated by other female (n=81). Six 3D body types were extracted using the cluster analysis. A : Slightly-slim & feminine image, B : Y-20's Ideal & Underweight figure image, C : E-70's Small-bust, Stooped back & large stomach figure image, D : Overweight figure image, E : Slightly- overweight & Feminine figure image, F : M- 40's slightly wide, Square shoulder & Masculine figure image. 4) We selected the image key words for making the 3D garment simulations. First, 136 image words were selected from 1075 words found in 136 Japanese ladies fashion books (2000 to 2004. Second). Those 136 words were categorized using five main image key word groups on a questionnaire (n=687, female respondents of 20's - 70's in age). Style 1 ; Lady & Feminine Image, Style 2 ; Sharp & Modern Image, Style 3 ; Casual Image Style, Style 4 ; Standard & Retro (old-fashioned) Image, Style 5 ; Elegant Image Style. 5) The same 20 design garment images and the decision making factors for labeling the design images were simulated on three 3D-body models (B, C, and F type 20's- 70's in age of 3) and each image displayed of the computer screen. Each 3D simulation was evaluated (n=120, female of 20's - 80's in age). The design image factors were categorized using the five main image key word and 38 design image items and 35 lifestyle items for each garment. Style 1: Skirt suit and pink color texture, etc., Style 2 : Black trouser suit of skin-like texture, etc., Style 3 : T-shirts of sky blue cotton-like texture & jeans of dark blue, Style 4 : One-piece dresses of soft textures, Style 5 : Skirt suit of white lace cotton-like texture and Suit dress etc.. Age differences of the evaluators produced few variations in the evaluation of the garment image. Less