基于ARShadowGAN的增强现实场景中物体阴影生成算法——以苏州博物馆藏品为例

Object shadow generation in augmented reality scenes based on ARShadowGAN:taking Suzhou Museum’s Collection as an example

在增强现实应用中,虚拟物体的阴影对真实性起着重要作用,但通常难以获得逼真的虚拟阴影。文章实验了一种基于生成对抗网络的用于增强现实场景中物体阴影生成的算法,并以苏州博物馆的珍贵藏品为例进行了详细研究。ARShadowGAN算法采用生成对抗网络的方法,通过深度学习生成合理的虚拟物体阴影,以更好地融合虚拟物体与真实场景。文章以苏州博物馆的藏品为案例展开研究,证明该算法的有效性和适用性。实验结果表明:ARShadowGAN能在增强现实场景中生成逼真的物体阴影,使虚拟物体与真实世界无缝融合。

In augmented reality applications,the shadows of virtual objects play an important role in realism,but it is often difficult to obtain realistic virtual shadows.The paper experimented with an algorithm based on generative adversarial networks for generating object shadows in augmented reality scenes,and conducted a detailed study taking precious collections from Suzhou Museum as an example.The ARShadowGAN algorithm uses the method of generating adversarial networks to generate reasonable virtual object shadows through deep learning,in order to better integrate virtual objects with real scenes.Taking the collections of Suzhou Museum as an example,demonstrate the effectiveness and applicability of the algorithm.The experimental results show that ARShadowGAN can generate realistic object shadows in augmented reality scenes,seamlessly integrating virtual objects with the real world.