Page 31 - GCN, June/july 2017

EMERGING TECH
BY PATRICK MARSHALL
Gigapixel image analysis on the fly
MOST FARMERS probably never thought they’d
need a way to process
huge digital images more quickly — until inexpensive drones with high-resolution cameras gave them access to tools they could use to micromanage irrigation and detect the growth of crop- threatening diseases.
Farmers usually send
a drone operator to scan the fields, bring all the imagery back to the office, download it to a computer and then upload it to the cloud. “So you’re 24 hours out before you get to
see your data,” said Amy Gooch, chief operations officer at ViSUS. “We are trying to make it so that you can see your data right there in the field.”
A spinoff from the University of Utah’s Scientific Computing
and Imaging Institute, ViSUS — which stands
for “visualization streams for ultimate scalability”
— is developing data- streaming techniques for progressively processing and displaying ultra-large image files, which include drone-generated scans of crops and visualizations of the complex connections between neurons in the brain.
The company’s ViSOAR Ag Explorer stitches the drone images together
to generate a map, then renders it for an iPhone, iPad or computer monitor. “It only renders the amount of data that you have the screen space for,” Gooch said. “It is progressively
real time on a variety of commercial platforms. The company said ViSUS has been deployed in a variety of large data applications, including monitoring scientific simulations and
of increasing resolution
as the user clicks to see more details. “It’s a little like Google Maps, where you zoom in and in, except with our technique, you don’t reload all the data each time,” said Valerio Pascucci, a computer science professor at the SCI Institute, where the technology was developed. “You only download the differences between levels, so it’s faster and requires less memory.”
The ViSOAR Ag Explorer team is also developing analytical tools, including multispectral and time- series processing that allows for pixel-value differencing over time. According to the team,
the combination of quick imagery access and analytical tools will enable farmers to optimize the
use of fertilizers, pesticides and herbicides, and spot potential problems, such as diseased plants or standing water.
ViSOAR Ag Explorer is being developed with the help of a $150,000 grant from the National Science Foundation’s Small Business Innovation Research Program. According to Gooch, the team expects
to release a beta version of ViSOAR Ag Explorer within a year and a final version within two years. •
“It only renders the amount of data that you have the screen space for. It is progressively refining, and as you zoom in, it goes and fetches more data.”
AMY GOOCH, VISUS
The ViSUS data-streaming application can progressively process and display ultra-large image files in real time on consumer hardware.
refining, and as you zoom in, it goes and fetches more data.”
ViSUS stores the images in a raw data format that enables efficient, streaming pipelines that process
the information while
it’s in motion. Users can manage large datasets in
editing massive images and panoramas.
The ViSOAR app allows users to browse gigapixel images, annotate them and adjust the imagery’s brightness, contrast, hue and saturation on the fly.
The tool is hierarchical, meaning it loads images
GCN JUNE/JULY 2017 • GCN.COM 27