Cascade classifier clears up clutter from radar

A technology that started life in detecting faces and penguin markings in the Antarctic is being adapted to help aid operators of military radar screens.

Cascade classifier clears up clutter from radar
Cascade classifier clears up clutter from radar

A technology that started life in detecting faces and penguin markings in the Antarctic is being adapted to help aid operators of military radar screens.

Engineering consultancy Frazer-Nash (Plymouth, UK) has been working on a project for the UK Defence Science Technology Laboratory (DSTL; Porton Down, UK) to explore the benefits of using a mathematical technique called a ‘cascade classifier’ on standard synthetic-aperture radar (SAR) images commonly used in the military.

This can help detect relevant objects on a radar screen and feed this information back to operators. In turn, this could help improve operator concentration and reduce potential stress caused by analyzing irrelevant information.

Crucially, the technology - which can be used on any existing SAR system - analyzes radar information in real time and can be used in many different environments, both static and mobile. It could become a cost-effective addition to military radar systems in the future once testing has been completed later in 2012.

“At any point while working in the field, a radar operator can be met with a large amount of data on a radar screen that needs to be quickly analyzed, and a decision made as a result. What the cascade classifier brings to operators within the UK Ministry of Defence is an aid to more effectively filter out unnecessary information that they encounter,” says Sam Dugelay at Frazer-Nash, who was responsible for leading the project.

Conservationists working in the Arctic have been successfully using a similar technique for many years to identify individual penguins from within a large colony. What the Frazer-Nash researchers have done is to look at how the technique could be repurposed for use in defense to help radar operators prioritize more effectively.

Readers interested in cascade classifiers might like to view a tutorial available here.

-- by Dave Wilson, Senior Editor, Vision Systems Design

More in Home