Research On Product Customized Modeling Design Method Driven By Multiple Images

Nowadays,the design concept of "people-oriented" and "user-centered" is more and more popular with the development of industrial design,and the product design issues that designers are concerned about more satisfying the emotional needs of consumers and personalized aesthetics.Personalized customization,as a design model that has emerged rapidly in recent years and it has been highly concerned by enterprises and users.This model pays more attention to the user’s personalized needs and user experience,giving the user great freedom in the initial design process of the product,but it is more reflected in the product function constraints,color schemes and module combinations,etc.It still needs to be improved in the Form modeling.At the same time,the appearance of the product largely reflects the overall emotional image of the product,and the current product modeling design still largely relies on the designer’s own design experience and creative thinking,which makes the product modeling image style unstable to some extent.Based on this,this study proposes to combine product customization and Intelligent design of product image modeling,,and build product modeling image mapping model,product modeling The Evaluation Model of Image and product image modeling intelligence based on perceptual engineering,artificial neural network and multi-objective optimization algorithm.It needs to establish a multi-image-driven personalized product modeling customization system to improve the user experience of personalized product modeling from the aspect of personalized product image modeling.The main research content includes the following three aspects.Firstly,the research needs the product image mining.It needs to collect and filter out multiple representative image vocabulary and a certain number of research samples after determining the research object,establish a mapping model between product modeling and image.Meanwhile,the curve control method is used to parametrically study sample modeling,so that a knowledge base of modeling images necessary for system implementation can be constructed.Secondly,it needs to establish a product modeling The Evaluation Model of Image.Based on the modeling image knowledge base,The Evaluation Model of Image was constructed based on BP neural network and convolutional neural network respectively.After comparing and analyzing the accuracy,error and performance of the two,it is found that The Evaluation Model of Image based on the convolutional neural network has smaller errors and higher accuracy,so this model is used to calculate the fitness of the new sample in the intelligent modeling design process.Thirdly,it needs to build an intelligent design model of product image modeling based on multi-objective particle swarm optimization（MOPSO）.The parameterized initial research sample is used as the initial particle population of the algorithm,three image words are selected as the target image at the same time,and related parameters are set,and The Evaluation Model of Image is used to perform intelligent calculation of particle fitness,and product modeling driven by multiple images Intelligent design;it also needs to select another three image vocabulary as the target image to carry out the intelligent design of multi-image driven modeling again,to obtain completely different product intelligent design results of three multi-objective optimization results,and to achieve personalized customization of product modeling.In this research,the liquor packaging bottle is used as a practical research case to design a personalized customization system for product styling driven by multiple images.It deeply explores the user’s image recognition and product modeling elements,establishes intelligent evaluation model and design model,and also establishes user-oriented personalized customization system of product modeling.It can satisfy the users’ personalized product modeling needs from the aspect of image modeling to some extent.