艺术风格智能认知辨识及其机械设计应用研究

Research on Intelligent Recognition of Art Design Style Based on Human Cognitive Mechanism and its Application in Mechanical Aesthetic Design

机械产品美学设计是机械制造中的一个重要组成部分,而设计艺术风格的界定是美学设计中的关键,它主要是人类认知机理的反映,但目前对艺术风格的识别算法多限于将艺术风格进行量化,忽略了人类感知在设计艺术风格分类中的重要作用,存在特征筛选困难、算法内在机制解释性差等问题。对此,创新性提出了一种基于人类认知机理的设计艺术风格智能识别方法。依托深度卷积神经网络模型构建了ACNN网络主线,对设计图像进行了特征提取,通过三种可视化方法对各卷积层进行特征可视化,结合人类对各层特征图的认知调整网络参数,并融入人类认知辅助结构,基于人类认知机理驱动网络,使网络能够重点筛选有效特征,最后通过实验进一步验证了此方法的有效性。研究是设计艺术风格识别方法研究领域的有益补充,可为机械产品提供美学设计指导。

The aesthetic design of mechanical products is an important part of mechanical manufacturing. The recognition of design art style is the key in aesthetic design,it is mainly the reflection of human cognitive mechanism. However,the recognition algorithm of art style is limited to quantify the artistic style,neglecting the important role of human perception in the classification of design art style,which has the difficulty of feature selection and poor interpretability of internal mechanism. In this paper,an intelligent classification method of design art style based on human cognitive mechanism is proposed. Based on the deep convolution neural network model,ACNN network mainline is constructed,and the image feature is extracted. The features of each layer are visualized by three visualization methods. The network parameters are adjusted according to the cognition of human beings on each layer of feature map,and the auxiliary structure of human cognition is integrated. Based on the human cognitive mechanism,the network can focus on screening effective features,Finally,the effectiveness of the method is further verified by experiments. Research is a useful supplement to the field of design art style recognition method,which can provide aesthetic design guidance for mechanical products.