【作者】 张全；

【导师】 陆长德；

【作者基本信息】 西北工业大学 ， 机械制造及其自动化， 2007， 博士

【摘要】 色彩是影响消费者选择并购买产品的重要因素，然而产品色彩设计一直是工业设计创新的薄弱环节，很多产品因为色彩不美观、不宜人而滞销。产品色彩智能设计(PCID)技术是在传统产品色彩设计理论基础上结合先进CAD技术和人工智能理论来解决色彩设计创新问题的新途径，目前面临的主要问题是：尚未建立健全的计算机智能思维体系；缺乏表征色彩设计意念、形象以及意象的有效方法；缺少适用于智能化色彩设计与评价的系统理论与方法。论文依托国家自然科学基金项目(编号：50275122)和国家863高技术研究发展计划项目(项目1编号：2002AA411120，项目2编号：2002AA411110)，从基础理论到工程应用对产品色彩设计的智能化问题进行了系统深入的研究，综合运用心物实验理论、色彩学、计算机辅助设计技术以及人工智能理论提出了一套基于复合思维的产品色彩智能设计理论方法，主要包括复合思维体系框架、交互式色彩调和技术、设计意象捕捉与评估技术、形色耦合美度评价方法以及产品色彩方案的自动生成与优化技术，具体内容如下：(1)产品色彩智能设计的复合思维理论体系框架人类设计活动的本质是思维，研究设计思维对于解决产品创新问题具有重要价值。与公理设计、可拓设计、TRIZ等传统设计理论不同，论文根据现有思维科学研究成果提出了复合思维概念和基于复合思维的产品色彩智能设计(PCIDBMT)技术。结合可拓理论建立了复合思维的对象模型、操作模型以及关系模型，构建了PCID的复合思维体系框架从而为PCIDBMT技术研究奠定理论基础。(2)交互式色彩调和技术传统CAD系统缺乏符合视觉规律的色彩形象模型和色彩调和工具。论文引入真正符合视觉规律的Munsell色彩体系，采用神经网络技术构建了面向形象思维的色彩调和模型(CHMOIT)，将CHMOIT与Photoshop系统中的均匀色彩空间Lab进行比较，结果表明CHMOIT具有更好的视觉秩序与均匀性。在CHMOIT基础上提出了4类交互式色彩调和控制算法，这些算法能根据设计师选择的主色调与调和类型快速生成对应的系列调和色，从而把设计师从繁琐的色彩调和工作中解脱出来。(3)色彩设计意象捕捉与评估技术根据复合思维理论引入了色彩意念、色彩形象和色彩意象的概念，提出了基于多维语义的色彩表征技术，利用集合论、测度论阐明了色彩形象与意念语义的度量特征，探讨了色彩形象与色彩意念空间映射特点，提出了消除由语义维度与语义粒度冲突而引起语义空洞的算法。针对设计意念的主观性问题提出了意象可信度评估技术，构建了基于粗集理论的意象可信度评估模型，为产品色彩设计定位提供定量评价技术。(4)形色耦合美度评价(SCCA)方法传统美学法则与蒙一斯潘赛(MS)色彩调和理论均不能定量评价复杂色彩方案的美度。论文在伯克霍夫美度模型、MS色彩调和理论以及心物场理论基础上提出形色耦合美度评价方法，基于图论提出了形色耦合拓扑图方法并用于表达形态与色彩的调和关系，基于视觉注意机制建立了多视点形色耦合美度模型，实验表明该方法的评价结果与专家评价结果基本一致。(5)基于遗传与免疫理论的产品色彩方案生成与优化技术把基因遗传和生物免疫理论引入产品色彩智能设计中，提出以产品组件为基本进化单元的基因编码方法，根据形色耦合美度模型构造了色彩方案适应度函数，综合应用遗传算法与免疫算法模拟色彩设计复合思维的发散和聚敛过程并构建了色彩创新模型，算例表明该模型能解决产品色彩方案的自动生成与优化问题。最后，综合应用上述理论方法开发了PCIDBMT原型系统。以一款手机产品为例对系统进行了测试，结果表明该系统能根据设计师的设计意象在短时间生成一些美观、宜人的产品色彩方案，从而验证了PCIDBMT理论与方法的有效性和可行性。更多还原

【Abstract】 Product color is one of the most important factors attracting consumers to select andbuy a product. However, product color has been the weakness of industrial design innovationso far. Many products are unmarketable for their unpleasant and ugly colors. Nowproduct-color intelligent design (PCID) technique is rounding into a new approach to solvethe problem based on conventional product color design theories, advanced CAD techniqueand artificial intelligence (AI) theory. Now PCID study is confronted with several keyproblems: a sane computer intelligent thinking system need to be established; no effectivemethod to represent color design concept, imagery and concept-image; no practicablesystemic theories or methods for product-color intelligent design and evaluation.With the sponsor of national natural science foundation (NO: 50275122) and national863 project (NO: 2002AA411120&2002AA411110), this paper studies the key techniques ofPCID from basic theory to engineering application. Applying psycho-physical theory, colorscience, CAD and AI theory, it builds a new technique called product-color intelligent designbased on multi-thinking (PCIDBMT), including multi-thinking (MT) system framework,interactive color harmony, design concept capture and evaluation, shape-color couplingaesthetics (SCCA) estimate, automatic design and optimization of product color schemes.The primary contents of this paper are as follows:(1) An MT system framework has been constructed for PCIDThe essence of design activity is thinking, therefore the study of design thinking theoryis a significant way to achieve product innovation. Based on the overview of current researchin thinking science, this paper presents the MT concept and a new design method calledPCIDBMT which is different from the conventional design theories such as axiomatic designtheory, extension design method and TRIZ technique,.etc. According to extension theory, itconstructs a concept model, an object model, an operation model and a relationship model ofMT. It also builds an MT system framework for PCID according to product color designprocess, and this framework set up a basic theory for PCID technique study.(2) Interactive color harmony techniquesConventional CAD systems lack of color harmony tool and color image model whichare in accord with vision rule. This paper introduces Munsell color system which is really inaccord with vision rule, and then it builds a color harmony model orient image thinking(CHMOIT) based on ANN technology. A contrast experiment has been done betweenCHMOIT and the Lab model in Photoshop system. Experiment results indicate thatCHMOIT has better vision order and uniformity. Four interactive color harmony controlalgorithms have been set up based on CHMOIT, which can produce series harmony coloraccording to the primary color and harmony type selected by designer. Hence designers mayextricate from the fussy work of color harmony.(3) Color design concept-image capture and evaluation methods This paper introduces color concept, color imagery and color concept-image into PCIDstudy according to MT theory. It sets up a representation method based on multidimensionalsemantic space, analyses the measurement features of color image and semantic conceptwith set theory and measure theory, discusses the mapping features between color image andcolor concept spaces, and presents an algorithm to avoid semantic hole brought by theinterference between semantic dimension and granularity. Further more, this paper putsforward a concept-image reliability estimate technique as for the problem of color concept’ssubjectivity, constructs a reliability evaluation model based on rough set theory and alsoprovides a quantitative evaluation technique for product color design orientation.(4) Shape-color coupling aesthetics (SCCA) evaluation methodsNeither Conventional aesthetics principles nor Moon-Spencer (MS) theory can evaluateproduct appearance’s aesthetics quantificationally. Based on G. Birkhoff aesthetics model,MS theory and psycho-physical field theory, this paper puts forward SCCA evaluationmethod. It applies shape-color coupling topological map to representing the relationshipbetween color and shape,, and constructs a multi-focus SCCA model based on visionattention mechanism. Psycho-physical experiments indicate that the evaluation results of thismethod are in accord with the results offered by experts.(5) Product color intelligent design and optimization techniques based on genetic andimmunity theoryThis paper introduces genetic and immunity theory into PCID study, establishes a genecoding principle based on product subassembly, constructs a fitness function of colorschemes according to SCCA model, applies genetic and immunity algorithm to simulate theradiation and convergence process of MT, at last it builds a product color design model.Experiments prove that this method can design and optimize color schemes on its own.Finally, it develops PCIDBMT system integrating theories and methods above. Amobile color design process in PCIDBMT system is tested and discussed in brief. The testshows PCIDBMT system can produce a mass of beautiful and pleasant schemes in a shorttime according to design concept, which proves that the theories and methods of PCIDBMTare feasible and effective.更多还原