Emergent Generation of Camouflage Patterns by Artificial Ants
Emergent Generation of Camouflage Patterns by Artificial Ants
复制标题
DOI：
10.1109/taai57707.2022.00040
复制
发表时间：
2022-12
期刊：
影响因子：
--
通讯作者：
Koichiro Sato;Kazunori Mizuno
中科院分区：
文献类型：
--
作者：
Koichiro Sato;Kazunori Mizuno
迷彩图案是一种纹理图案，包含各种颜色和局部形状，与风景或物体相融合。虽然大多数伪装图案通常是基于人类主观性通过随机组合颜色和各种局部形状来创建的，但是根据特定的背景图像生成合适的伪装图案可能是困难的。在本文中，我们提出了一种方法，可以紧急生成复杂的伪装图案，通过使用蚂蚁绘画算法。在该方法中，在输入图像的一些特征被提取后，人工蚂蚁根据特征绘制伪装。我们证明，通过蚂蚁之间的相互作用，所提出的方法可以生成更复杂的伪装图案，但更好地融合与背景图像比随机生成。
A camouflage pattern is a texture pattern that contains various colors and localized shapes to be blended with landscapes or objects. Although most of camouflage patterns are generally created by randomly combining colors and various local shapes based on human subjectivity, it can be difficult to generate an appropriate camouflage pattern according to a specific background image. In this paper, we propose a method that can emergently generate complex camouflage patterns by using an Ant Painting algorithm. In the proposed method, after some features of an input image are extracted, artificial ants draws a camouflage according to the features. We demonstrate that, by interaction among ants, the proposed method can generate camouflage patterns which are more complex but blends better with the background image than random generation.