817

Robust Real Time Pattern Matching

Werman Michael, HUJI, School of Computer Science and Engineering, Computer Science
Pele Ofir, HUJI, School of Computer Science and Engineering

tamir_huberman_-_michael_werman.jpg

Novel solution for automatic robust, real-time pattern matching

Categories

Computer science, imaging, computer graphics

Development Stage

Product ready for commercialization; ongoing algorithm improvement

 

  

Highlights

  • Proprietary algorithm for pattern matching in both static and dynamic applications
  • Computer vision software for extracting and analyzing information from images
  • Simple, fast method for use by industrial robots and vision based systems for assembly, manufacturing and inspection
  • Applications in face recognition and security

Our Innovation

  • Two part system
  • The “Image Hamming Distance Family” of image distance measures. Family members are robust to occlusions, geometrical transforms, light changes and non-rigid deformations.
  • Novel Bayesian framework for sequential hypothesis testing on finite populations
  • Based on the framework, an optimal rejection/acceptance sampling algorithm is designed. The algorithm quickly determines whether two images are similar with respect to a member of the Image Hamming Distance Family.

Key Features

  • On a Pentium 4 3GHz processor, detection of a pattern with 2197 pixels in 640x480 pixel frames, where in each frame the pattern rotated and was highly occluded, proceeded at only 0.022 seconds per frame
  • Enables searching for certain parts in an image or aligning parts to a reference position

The Opportunity

  • Applications in imaging/video editing and enhancement - nearly 50% of digital cameras buyers use software for managing and enhancing their photographs
  • For serious amateur and professional mphotographers, postproduction (any possible manipulation) is essential. They use new, easy-to-use programs (to correct red eyes, face recognition, scene recognition and tonal corrections) in addition to improved in-camera algorithms and many new programs to produce high quality pictures in minilabs
Additional Website: http://www.cs.huji.ac.il/~ofirpele/hs/

Patent Status

Granted US 8,417,655

Contact for more information:

Aviv Shoher
SVP BUSINESS DEVELOPMENT
+972-2-6586635
Contact ME:
Image CAPTCHA