基于交互式遗传算法的服装几何花卉图案的设计

发布时间：2018-06-19 18:47

本文选题：服装 + 几何花卉图案　；参考：《浙江理工大学》2017年硕士论文

【摘要】：作为服装图案的重要组成元素,几何、花卉图案历经千年,表现上虽千变万化,但是作为承载美,展现人类精神风貌、表达文化诉求的重要载体,至今仍然是构成服装图案的重要元素之一,数码印花技术的开发与应用,也使服装的图案的呈现形式不再过度依赖于缝合、绣花工艺。因此,本文从智能算法的理论出发,依据服装几何花卉图案的排列构成规律,设计出服装几何花卉图案,丰富服装图案的设计方法。本文首先对交互设计和交互式遗传算法的相关理论进行分析研究,以交互设计的五要素为基础,进行交互界面的规范和设计。将交互式遗传算法的编码、选择、杂交、变异以及适应性评价等过程与传统服装几何花卉图案的设计流程相结合,针对交互式遗传算法容易产生的用户评价疲劳的问题,本文采用了两种解决办法:通过对传统交互式遗传算法流程进行改进,减少算法中原有次优方案的生成,从而减少用户在评价数量上的疲劳;通过引入径向基函数神经网络代理模型,对用户的相同评价行为进行代理,减轻用户在评价动作上的疲劳。随后,本文通过对服装几何、花卉图案的历史演变的研究,总结出几何花卉图案在服装面料上的布局规律,作为本文几何花卉图案交互设计程序的指导规则,使程序生成比较符合传统审美的几何花卉图案的搭配方案。本文还通过改变交互式遗传算法中的参数取值范围来实现设计方案中几何和花卉位置的变形,以生成更多不同排列形式的设计方案。最后通过数码印花实践证明本研究方法得出设计图案的可行性。本文的重点是将数学智能算法与服装图案设计过程相结合,找到二者相结合的切入点,生成较为有秩序并具有一定美观的服装图案设计方案。基于交互式遗传算法相关理论,在JAVA环境下进行交互设计程序的运行,通过实践证明,根据输入的图案素材,通过程序的进化迭代步骤,并使图案之间的平面位置发生变化,可以更加快速、大批量生成具有某种特定规律的一类服装图案。
[Abstract]:As an important component of clothing patterns, geometry and flower patterns have experienced thousands of years, although the performance of the ever-changing, but as a bearing of the United States, to show the spirit of the human face, an important carrier to express cultural aspirations, Up to now, it is still one of the important elements of garment pattern. With the development and application of digital printing technology, the presentation form of garment pattern is no longer dependent on suture and embroidery technology. Therefore, based on the theory of intelligent algorithm and the rule of arrangement and constitution of geometric flower pattern, this paper designs the geometric flower pattern of clothing, which enriches the design method of garment pattern. In this paper, the interaction design and interactive genetic algorithm related theories are analyzed and studied. Based on the five elements of interactive design, the interaction interface is standardized and designed. This paper combines the coding, selection, hybridization, mutation and adaptability evaluation of interactive genetic algorithm with the design process of traditional geometric flower pattern, aiming at the problem of user evaluation fatigue caused by interactive genetic algorithm. This paper adopts two solutions: by improving the traditional interactive genetic algorithm process, reducing the generation of the original sub-optimal scheme in the algorithm, so as to reduce the fatigue of users in the evaluation quantity; By introducing radial basis function (RBF) neural network agent model, the same evaluation behavior of users is represented and the fatigue of users in evaluating actions is alleviated. Then, through the study of the historical evolution of the geometric pattern and the pattern of flowers, this paper sums up the layout rule of the geometric flower pattern on the fabric, which is the guiding rule of the interactive design program of the geometric flower pattern in this paper. Make the program more in line with the traditional aesthetic geometric flower pattern matching scheme. This paper also changes the range of parameters in interactive genetic algorithm to realize the deformation of geometry and flower position in the design scheme, in order to generate more different arrangement forms of design scheme. Finally, the feasibility of the design method is proved by digital printing practice. The emphasis of this paper is to combine the mathematical intelligent algorithm with the process of garment pattern design, to find the breakthrough point of the combination of the two, and to produce a more orderly and beautiful clothing pattern design scheme. Based on the theory of interactive genetic algorithm, the interactive design program is run in Java environment. It is proved by practice that, according to the input pattern material, the evolutionary iterative steps of the program are adopted, and the plane position between patterns is changed. Can be faster, mass production of a certain rule of clothing patterns.
【学位授予单位】：浙江理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TS941.2

【参考文献】