HybridFlow Turns Aerial Imagery Into 3D "Digital Twins" in Half an Hour — with Zero Machine Learning
By sectioning images and finding similarities at a pixel level, HybridFlow is able to turn 2D photos into 3D point clouds fast.
A pair of researchers from Concordia University have created a tool for rapidly creating "digital twins" of a landscape — offering a faster, more accurate alternative to current scanning systems which works "down to the pixel level."
"This digital twin can be used in typical applications to navigate and explore different areas, as well as virtual tourism, games, films and so on," explains Charalambos Poullis, associate professor of computer science and software engineering and senior author of the work. "More importantly, there are very impactful applications that can simulate processes in a secure and digital way. So, it can be used by stakeholders and authorities to simulate 'what-if' scenarios in cases of flooding or other natural disasters. This allows us to make informed decisions and evaluate various risk-mitigating factors."
That digital twins — designed to represent, as accurately as possible, some real-world environment or object — are useful is not news, but creating them to any level of accuracy has always been a laborious task. It's this task that the team's creation, HybridFlow, aims to solve, using motion estimation to turn aerial imagery into precise 3D models.
The system works by clustering image segments depending on how similar they look, based on a pixel-level analysis. Points of interest are then trackable across images with less processing time and improved accuracy compared to existing approaches — to the point that an "average-sized model" of a built-up area could be produced from aerial imagery in under 30 minutes. "It also eliminates the need for any deep learning technique, which would require a lot of training and resources," Poullis adds of HybridFlow. "This is a data-driven method that can handle an arbitrarily large image set."
Poullis, with first author Qiao Chen, is currently working with the city of Terrebonne, northeast of Montreal, to use the HybridFlow system to create a digital twin for disaster planning. "They know they cannot prevent the flooding, but we can provide them with tools to make informed decisions," says Poullis. "We allow them to change the environment by introducing barriers such as sandbags, and then we run simulations to see how the floodwater flow is affected."
The team's work has been published in the journal Scientific Reports under open-access terms.
