基于深度表征学习和遗传算法的军用座舱色彩设计方法  

Military Cockpit Color Design Method Based on Deep Representation Learning and Genetic Algorithm

作　　者：苏胜[1] 顾森 宋志强 刘萍[1] SU Sheng;GU Sen;SONG Zhiqiang;LIU Ping(School of Art and Media,Xi an Technological University,Xi an 710021,Shaanxi,China;School of Mechanical and Electrical Engineering,Henan University of Technology,Zhengzhou 450001,Henan,China;Henan Industrial Design Institute,Henan University of Technology,Zhengzhou 450001,Henan,China;School of Art and Design,Xi an Fanyi University,Xi an 710105,Shaanxi,China)

机构地区：[1]西安工业大学艺术与传媒学院,陕西西安710021 [2]河南工业大学机电工程学院,河南郑州450001 [3]河南工业大学河南省工业设计研究院,河南郑州450001 [4]西安翻译学院艺术与设计学院,陕西西安710105

出　　处：《兵工学报》2024年第4期1060-1069,共10页Acta Armamentarii

基　　金：2020年教育部人文社会科学规划基金项目(20YJAZH092)。

摘　　要：军用座舱色彩设计作为载人军事设备工业设计中较为主观的一部分,设计的合理性至关重要。为提高军用座舱色彩设计的科学性,提出一种基于深度表征学习和遗传算法的军用座舱配色方法。利用深度表征学习模型预测军用座舱配色方案,并根据色彩感知理论建立军用座舱配色模型,将其作为生成方案的限制条件。同时,引入交互式遗传算法到智能配色系统中,通过人工引导的方式优化神经网络的参数,对预测的配色方案进行有效迭代。实验结果表明:该方法生成的配色方案符合军用座舱配色模型,结合遗传算法的模型预测准确率比单一的深度表征模型提高了16%~18%。相较于人工色彩设计方案,军用座舱智能配色方法生成的方案满意度略优、设计周期缩短了80%~88%,色彩稳定性提高了6%~12%。The color design of military cockpits is a subjective part of industrial design for manned military equipment,and the rationality of the design is crucial.A military cockpit color matching method based on deep representation learning and genetic algorithm is proposed to improve the scientific nature of military cockpit color design.This method uses a deep representation learning model to predict the military cockpit color schemes and establishes a military cockpit color matching model based on color perception theory,which serves as a constraint condition for generating schemes.At the same time,an interactive genetic algorithm is introduced into the intelligent color matching system to optimize the parameters of neural network through manual guidance,and effectively iterate the predicted color matching schemes.The results show that the color matching schemes generated by the proposed method comply with the military cockpit color matching model,and the prediction accuracy of the proposed model combined with genetic algorithm is improved by 16%~18%compared to a single deep representation model.Compared to manual color design schemes,the satisfaction of the color schemes generated by the military cockpit intelligent color matching method is slightly higher,the design cycle is shortened by 80%~88%,and the color stability is improved by 6%~12%.

关 键 词：军用座舱配色 深度表征学习 交互式遗传算法 色彩感知 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]