A new study by researchers from The University of Western Australia has used computer vision to reconstruct the false illusion of 3D shapes created by moth wing patterns.
The research, published in Royal Society Interface used computer vision, which is typically used for applications such as self-drive vehicles and face recognition, to determine whether the flat wing patterns of the green fruit-piercing moth, Eudocima salaminia, were falsely reconstructed as 3D shapes. This species of moth, which is commonly found across southeast Asia and Australia, is citrus fruit pest.
The work was led by Dr Jennifer Kelley from UWA's School of Biological Sciences, in collaboration with Professor Mohammed Bennamoun from UWA's School of Physics, Mathematics and Computing, Professor Farid Boussaid from UWA's School of Engineering, and former UWA computer science students Laurent Jospin and James Porter,
Dr Kelley said that the wing patterns were very convincing and made the moth look like a rolled leaf, which would help it avoid daytime predators such as birds.
"We used several state-of-the-art computer vision methods to reconstruct the visual illusions generated by the wing patterns," Dr Kelley said.
"Biologists have long thought that animal patterns could produce shape illusions, but finding evidence for this is difficult, because of the challenges associated with study visual perception in non-human animals."
The research revealed that many of the computer vision algorithms were fooled by the moth's wing patterns, with the reconstructions falsely producing 3D shapes.
"This is an unusual application of computer vision which may help us better understand the illusory effects generated by animal colouration," Dr Kelley said.
The researchers are now looking to discover whether these illusory effects occur in other animals, such as striped pattens in snakes, and for functions other than camouflage, including eyespots on peacock feathers that are used to attract mates.
