计算美学的生成与度量

The Generation and Measurement of Computational Aesthetics

计算美学作为一门跨学科研究领域,旨在探讨计算机科学与艺术的交汇点,让计算机得以像人类一样做出审美决策,包含美学的生成和度量两个方向,核心在于研究美的本质及如何利用计算机技术来实现和衡量美。计算美学的生成即利用计算机生成具备人类感知中美的特质的艺术作品,文章分析了当前美学生成研究核心的辅助生成、自动生成和风格迁移的现状与发展,讨论了美学生成的挑战与机遇。计算美学的度量即通过计算机技术实现对美的量化,文章从美学特征的视角介绍了美学度量的具体路径,以及生成式对抗网络、卷积神经网络等前沿的度量方法。

As an interdisciplinary research field,computational aesthetics aims to explore the intersection of computer science and art,so that computers can make aesthetic decisions such as aesthetic generation and measurement like human beings.The key is to study the essence of beauty and to use computer technology to create and measure beauty.The generation of computational aesthetics refers to the use of computer to generate art works with the characteristics of human perception.This paper analyzes the current status and development of assisted generation,automatic generation and style transfer,which are the core of aesthetic generation research,and discusses the challenges and opportunities of aesthetic generation.The measurement of computational aesthetics is to quantify beauty through computer technology.It introduces the specific path of aesthetic measurement from the perspective of aesthetic characteristics,and introduces the advanced measurement methods such as generative adversarial network and convolutional neural network.