Sunday, May 15, 2011

Image Based Search Engine Created - Finding A Whatchamacallit On The Web

April 1, 2007 — VizSeek is one of the first search engines on the Internet to use a photograph, a 2D image, or a 3D model and transform it into a 3D shape. The search can be narrowed with additional information. The image-seeking search engine produces search results in a matter of seconds.

Whether you need a bolt, a motor, a belt or a tool, finding the perfect, hard-to-describe part can be like finding a needle in a haystack. But now, engineers have put together a computer program that can track down just about anything you need -- even if you don't know what it's called!

When it comes to making repairs, replacing parts or finding a tool, a new visual search engine can help you find exactly what you're looking. Created by mechanical engineers, VizSeek uses pattern recognition to take a doodle, sketch or photo and find out exactly what it is.

"Every, every part has a unique fingerprint, and we call it shape representation," Nainesh Rathod, CEO of Imaginestics in West Lafayette, Ind., -- the company who developed VizSeek, tells DBIS.

VizSeek is one of the first search engines on the Internet to use a photograph, a 2D image, or a 3-D model and transform it into a 3-D shape to match the part you're searching for. You can help narrow the search with additional information. For example, tell VizSeek it's a part for a motor and it will sort search results based on what it found from thousands of motors, bolts, hinges and belts cataloged in the system.

"There's so much data that you're processing, and you have to do it in a matter of seconds," Rathod says. "Waiting a minute's not gonna do it."

Right now, VizSeek is used for manufacturers to track down parts or find suppliers but in the near future, it could help anyone track down just about any part they need.

Imaginestics also hopes to have stores catalog which parts they have. You would no longer have to take the part into the store and try and match it. Instead, you would just print out the barcode from your computer, take that in and have the salesperson find it for you.

You can access VizSeek at VizSeek.com.

BACKGROUND: A company in Purdue Research Park has created a new kind of search engine called 3D-Seek, thanks to a major advance in pattern recognition. The program lets users find hard-to-describe items -- hinges, bolts, conveyor belts, or motors, for example -- in an online catalog without ever needing to know the items -- names, part numbers or keywords. Instead, all a user needs to do is draw a simple freehand sketch : a doodle. The drawing can be done from any angle of the actual object. 3D-Seek will find the desired part in seconds via a related technology called i-prowler, which hunts for image files on a user's computer and merges them with either the online database or an internal company catalog

BENEFITS: There are differences in the ways various search engines work, but they all perform three basic tasks. They search the Internet -- or select pieces of the Internet -- based on important words. They keep an index of the words they find, and where they find them. They allow users to look for words or combinations of words found in that index. Instead of walking around from store to store carrying a bracket, for example, you can use 3D Seek to find a match online, print out a barcode for the exact part, and take it to any store that stocks it. An online test version is available at the moment, with a full-fledged product expected for launch in the fall.

ABOUT THE PROGRAM: The 3D-Seek software was built on top of technology created by the Purdue Research and Education Center for Information Systems in Engineering (PRECISE) at Purdue University, to compare computer-aided design files and other 3D images that are used throughout industry. Further collaboration led to a system that required only critical shape characteristics, not entire image files, which allowed faster search speeds and also protected proprietary information held by parts suppliers. The challenge was to match a rudimentary doodle to an actual 3D object in seconds.

The Human Factors and Ergonomics Society contributed to the information contained in the TV portion of this report.

Note: This story and accompanying video were originally produced for the American Institute of Physics series Discoveries and Breakthroughs in Science by Ivanhoe Broadcast News and are protected by copyright law. All rights reserved.
 
 

No comments: