NEC image-processing tech can detect invisible damage to bridges
- 09 December, 2014 20:12
NEC has developed a camera system that can detect deterioration in bridges and other structures simply by analyzing imagery.
The electronics manufacturer is calling it the world's first system to use high-definition video cameras and image-recognition technologies to analyze and estimate structural vibrations.
NEC researchers developed an algorithm that can pick up numerous minute vibrations on the surface of a structure by analyzing imagery of it, the company said on Tuesday. Image-compression know-how is used to speed up the analysis since the process generates large volumes of data.
The measurements are then processed with another algorithm to estimate the level of damage that may have taken place within a structure in terms of cracks, separations and hollow cavities. Invisible damage and wear can be estimated with a high degree of accuracy, according to the firm.
"In recent years, aging infrastructure has become an increasing concern, particularly civil engineering structures," a company spokesman said in an email. "NEC is aiming to provide technologies that promote safety and help maintain infrastructure well into the future."
In addition to bridges, the system could be used at construction sites, factories and other large-scale facilities where it's important to prevent down-time caused by maintenance work.
Although the system is still under development, it might be used with smartphones and other mobile devices in the future, the spokesman said. NEC is aiming to commercialize the system in the first half of 2016.
In 2011, the company came out with a high-precision piezoelectric vibration sensor that allows for real-time identification of structural damage including cracks in walls and gas and water lines.
More recently, NEC has used its research into image-processing techniques for a number of applications, ranging from smartphone camera technology that helps spot counterfeit goods to text-free magazines that can be "read" with an image-recognition app.