基于大数据图像处理的产品造型智能设计探索

Intelligent Design of Product Modeling Based on Big Data Image Processing

中文关键词: 数据驱动设计  智能设计  产品数据库  图像处理  生成对抗网络

英文关键词: Generation Method Based on Multimodal Neural Network[J]. Journal of Chinese Information Processing, 2017, 31(6):162-171.

中文摘要:

      目的 针对当前设计师囿于已有经验和当前的产品创新设计方法，导致设计师不能完全把握产品设计的流行趋势，易出现设计雷同和侵权等问题，提出了一种利用基于深度学习的图像处理等技术，研究基于产品图像数据库的智能造型设计方法，为设计师开发新产品提供灵感，优化产品设计流程。方法 首先利用爬虫技术从购物网站上抓取产品图像，利用图像处理算法去除重复、背景杂乱的图像。然后分析图像包含的客观特征，利用多标签学习技术对图像进行分类标注，形成一个可以实时更新、带有标签的多层次灵活分类的产品图像数据库，该数据库可以方便设计师利用标签检索。利用生成对抗网络技术进行风格迁移学习，生成与原图像类似但是有不同的新图像。最后经过有经验的设计师参与设计，得到新的方案草图。结论 基于数据驱动的造型智能设计方法，可以在一定程度上提高设计效率，缩短设计周期，为造型设计提供一种新的思维方式。

英文摘要:

      Designers now are constrained by their experience and the current product innovation design methods, which leads to designers can't fully grasp the trend of the current product design, and that it's easy to raise similar designs and infringement problems. Therefore, we used the big data image processing, deep learning and other technologies to propose an intelligent modeling design method based on product image database, which aims to provide inspiration for designers to develop new products and optimize product design process. Firstly, crawler technology is used to grab product images from shopping websites. The image processing algorithm is used to remove repetitive and cluttered images. Then the objective features contained in the image are analyzed. The multi-label learning technique is used to annotate the image. This way forms a multi-level, flexible and real-time-updated product image database with labels which is convenient for designers to search by labels. The deep learning image is made using the generative confrontation network technology to generate new images similar to the original image but different from the original one. Finally, through the participation of experienced designers in the design, the new scheme sketch was obtained. The data-driven modeling intelligent design method can improve design efficiency, shorten design cycle and provide a new way of thinking for designers.